{ "schemaVersion": 1, "kind": "AUTONOMY_ECONOMICS_PUBLIC_REGISTER", "register": "Autonomy Economics Evidence Monitor", "canonical": "https://autonomyeconomics.com/evidence", "frameworkPublished": "2026-04-30", "generatedAt": "2026-10-19T07:51:44.409Z", "licence": "Evidence records may be cited with attribution to the Autonomy Economics Evidence Monitor. Source material remains the property of its publishers and is linked, not reproduced.", "events": [ { "id": "AE-2026-046", "title": "Open decision-model alternatives emerge within weeks of Jev launch", "slug": "decision-model-capability-diffuses-within-weeks-of-jev-launch", "eventDate": "2026-10-01", "publicationDate": "2026-10-01", "discoveredDate": "2026-10-02", "datePrecision": "EXACT", "companies": [ "OpenAI", "TypeSafe AI", "Cloudflare", "Amazon Web Services", "Perplexity" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Within 16 days of TypeSafe AI launching Jev as a specialised System One decision model, Cloudflare and Strands released competing decision models with open weights or open-source code, while OpenAI announced a Decisions API and Perplexity made an Apache-2.0 decision model available. The cluster broadens access to a newly differentiated decision capability across hosted APIs, cloud platforms and local or self-hosted implementations.", "observedFact": "TypeSafe AI launched Jev on 15 September 2026 as its first public System One Model for fast, structured decisions. On 29 September OpenAI announced a Decisions API powered by Luna for classification, routing and choosing predefined actions. On 1 October Cloudflare launched Clef and Clef-flash, described them as decision models in the same family as Jev, made them available on Workers AI, open-sourced the weights under Apache 2.0 and made the System One API compatible enough for a Jev integration to switch by changing endpoint and model. Also on 1 October, Strands released the open-source Strands Decider 2B with weights, training data and scripts, suitable for local CPU or GPU use. Perplexity also made an Apache-2.0 decision model, pplx-decider-v1-27b, available with downloadable weights.", "interpretation": "This is strong early supportive evidence for Capability Scarcity Dependence. A capability that initially appeared as a differentiated specialist model category was replicated and made substitutable across several providers within weeks, including through open weights and local deployment. The CSD signal is the rapid broadening of access to the underlying decision capability. It does not claim that Jev has lost all commercial advantage or that scarcity-dependent rents have weakened.", "mechanisms": [ "CSD" ], "laws": [], "propositions": [], "themes": [ "capability-scarcity", "decision-models", "open-weights", "model-substitution", "open-source" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "horizonNote": "This is a current capability-access signal. The longer-run question is whether these alternatives become credible production substitutes and whether commercial differentiation migrates toward data, fine-tuning, integration, workflow ownership, distribution, governance, calibration or orchestration.", "predictionTested": "Whether a newly scarce technical capability can be replicated, substituted and accessed through competing providers, cloud platforms and open weights, weakening scarcity around access to the capability itself.", "whyItMatters": "The sequence closely matches the mechanism described by CSD, in which value is exposed when a capability that was difficult to access, replicate or substitute becomes available through multiple credible routes. The speed of diffusion is particularly informative because the relevant capability category moved from a specialist launch to competing hosted and open implementations within weeks.", "counterInterpretation": "Jev may still retain superior calibration, accuracy, reliability, developer mindshare, hosting convenience or production performance. The alternatives may not be commercially equivalent at this stage, and scarcity can persist at the frontier even while lower-cost or open substitutes broaden access. The observation therefore shows weakening scarcity around access to the capability category. It does not show the disappearance of differentiation.", "whatWouldStrengthen": "Independent evidence that firms are switching between decision-model providers, self-hosting open alternatives, or seeing decision-model pricing converge would strengthen the CSD reading. Comparable production benchmarks and sustained use of the open models would also make substitution more economically credible.", "whatWouldWeaken": "Evidence that the competing models fail in production, cannot match Jev on consequential workloads, or that durable proprietary data, calibration, reliability or integration advantages keep the capability effectively non-substitutable would weaken the interpretation.", "falsificationRelevance": "The current observation supports the access, replication and substitution mechanism directly. At this stage it does not satisfy the paper's one-to-two-year commercial-unit durability test, because no sustained effect on scarcity-dependent pricing, margins or rents has been observed.", "relatedEvidenceIds": [], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-10-02T15:57:18.803Z", "createdAt": "2026-10-02T15:55:47.091Z", "updatedAt": "2026-10-11T09:24:49.873Z", "sources": [ { "id": "fd96d1f0-dfe6-4056-9d04-7a252963e128", "evidenceEventId": "AE-2026-046", "sourceName": "TypeSafe AI", "headlineAtVerification": "Introducing System One Models & Jev", "sourceUrl": "https://typesafe.ai/blog/introducing-system-one-models-and-jev", "canonicalUrl": "https://typesafe.ai/blog/introducing-system-one-models-and-jev", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-15", "retrievedAt": "2026-10-02T15:56:29.384Z", "lastVerifiedAt": "2026-10-02T15:56:29.384Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Official TypeSafe AI launch post dated 15 September 2026. It describes Jev as TypeSafe AI's first public System One Model for fast, structured decisions." }, { "id": "88e75cdd-8325-43dc-bda6-c47674e6f9d0", "evidenceEventId": "AE-2026-046", "sourceName": "OpenAI Developer Community", "headlineAtVerification": "DevDay 2026 announcements and developer resources", "sourceUrl": "https://community.openai.com/t/devday-2026-announcements-and-developer-resources/1402006", "canonicalUrl": "https://community.openai.com/t/devday-2026-announcements-and-developer-resources/1402006", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-29", "retrievedAt": "2026-10-02T15:56:33.821Z", "lastVerifiedAt": "2026-10-02T15:56:33.821Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Official OpenAI Developer Community announcements post dated 29 September 2026. It states that the Decisions API is in limited preview and uses Luna to classify inputs, route requests or choose an action from predefined answers." }, { "id": "2d358ba3-c634-4d1a-9aae-e4a8ffb68d7c", "evidenceEventId": "AE-2026-046", "sourceName": "Cloudflare", "headlineAtVerification": "Introducing Clef: our open-source decision models, and new RL fine-tuning platform", "sourceUrl": "https://blog.cloudflare.com/clef-decision-models/", "canonicalUrl": "https://blog.cloudflare.com/clef-decision-models/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-10-01", "retrievedAt": "2026-10-02T15:56:36.275Z", "lastVerifiedAt": "2026-10-02T15:56:36.275Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Official Cloudflare launch post dated 1 October 2026. It describes Clef and Clef-flash as Jev-API-compatible decision models and states that the models are open-sourced under Apache 2.0." }, { "id": "8181e7e0-d4ad-4698-b5c7-b8e400963a48", "evidenceEventId": "AE-2026-046", "sourceName": "Strands Agents", "headlineAtVerification": "Introducing Strands Decider 2B: a small, open source, decision model", "sourceUrl": "https://strandsagents.com/blog/introducing-strands-decider/", "canonicalUrl": "https://strandsagents.com/blog/introducing-strands-decider/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-10-01", "retrievedAt": "2026-10-02T15:56:39.014Z", "lastVerifiedAt": "2026-10-02T15:56:39.014Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Official Strands Agents post dated 1 October 2026. It describes Strands Decider 2B as a 2B-parameter open-source decision model suitable for local CPU or GPU use, with weights, training data and scripts released." }, { "id": "d6b81f0f-3a5c-4c54-9e8d-69407db2d3e5", "evidenceEventId": "AE-2026-046", "sourceName": "Perplexity", "headlineAtVerification": "pplx-decider-v1-27b", "sourceUrl": "https://huggingface.co/perplexity-ai/pplx-decider-v1-27b", "canonicalUrl": "https://huggingface.co/perplexity-ai/pplx-decider-v1-27b", "sourceType": "PRIMARY", "sourceTier": 1, "retrievedAt": "2026-10-02T15:56:42.583Z", "lastVerifiedAt": "2026-10-02T15:56:42.583Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Official Perplexity Hugging Face model card. It identifies pplx-decider-v1-27b as a decision model and lists the licence as Apache 2.0. The retrieved page did not expose a publication date, so none is asserted here." } ] }, { "id": "AE-2026-035", "title": "Blackstone links realised AI productivity gains to early repricing and a shift from seats to outcomes", "slug": "blackstone-ai-productivity-repricing-seats-to-outcomes", "eventDate": "2026-09-20", "publicationDate": "2026-09-20", "discoveredDate": "2026-09-29", "datePrecision": "OBSERVED_BY", "companies": [ "Blackstone" ], "sectors": [ "software", "consulting", "private-equity" ], "sectorLabels": [ "software", "consulting", "private-equity" ], "summary": "In a September 2026 investor presentation, Blackstone President and COO Jon Gray described substantial realised AI productivity gains across portfolio companies, then highlighted multiple compression in professional services, software and information-services businesses even where underlying operations remained sound, and said management teams that move the model from seats to outcomes should be better positioned.", "observedFact": "Blackstone's Jon Gray gave examples of realised AI gains including leases processed five times faster, an AI process generating $4.5 million a year after a $4 million investment, an 18x return on model spend in software engineering, customer applications processed 90% faster, and a Blackstone portfolio-intelligence workflow with timelines reduced by 99%. Later in the same presentation, he said stocks of professional services, software and information-services companies had fallen and that, in many cases, the underlying business was fine but valuation multiples had compressed because investors were uncertain about future business economics. He also said private-equity software deal activity had fallen 66% and argued that management teams that change the model from seats to outcomes would be better positioned.", "interpretation": "The presentation is notable less for introducing outcome-based pricing, which was already visible before September 2026, than for connecting realised AI productivity, valuation pressure on exposed business models and commercial-model redesign within the same institutional investor analysis. Its seat-to-outcome discussion is consistent with Law 2, while the wider presentation is more directly relevant to the Agentic Profit Paradox and Revenue Durability proposition.", "mechanisms": [ "HBP" ], "laws": [ "LAW_2" ], "propositions": [ "AGENTIC_PROFIT_PARADOX", "REVENUE_DURABILITY", "COMMERCIAL_ALIGNMENT" ], "themes": [ "outcome-pricing", "valuation-repricing", "early-pressure", "productivity" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "Whether autonomous capability can generate substantial operating gains while markets simultaneously question the durability of legacy commercial structures, and whether the proposed response shifts value capture away from seat-based access toward outcomes.", "whyItMatters": "It combines three parts of the proposed mechanism in one post-publication primary-source presentation, namely realised productivity gains, repricing before broad operating deterioration, and an explicit recommendation to move from seats to outcomes.", "counterInterpretation": "The productivity examples and the repriced public companies are not necessarily the same businesses, so the presentation does not establish a firm-level causal chain from AI productivity to valuation compression. Multiple compression and lower software deal activity can also reflect interest rates, macro conditions or ordinary technology-cycle uncertainty. The seat-to-outcome statement is a forward-looking management view and does not show that such pricing has already been adopted at scale.", "whatWouldStrengthen": "Named companies showing sustained AI-driven productivity gains followed by measurable seat contraction or pricing redesign, with outcome-based revenue preserving or improving revenue durability and valuation relative to comparable firms.", "whatWouldWeaken": "Evidence that exposed firms retain durable seat-based economics despite sustained autonomous productivity gains, or that the observed valuation compression reverses without any commercial-model redesign.", "falsificationRelevance": "This observation is supportive but not causal. It would weaken if exposed firms retain durable seat economics despite sustained autonomous productivity gains, or if repricing reverses without commercial redesign.", "relatedEvidenceIds": [ "AE-2026-019", "AE-2026-021" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-29T13:38:07.599Z", "createdAt": "2026-09-29T13:38:07.599Z", "updatedAt": "2026-10-11T09:24:49.396Z", "sources": [ { "id": "51ce85fc-dd69-40b4-a284-8532c52c9a0a", "evidenceEventId": "AE-2026-035", "sourceName": "Blackstone", "headlineAtVerification": "Where's the Beef? Blackstone's Jon Gray on the Payoff From the Enormous AI CapEx Spend | Sept '26", "sourceUrl": "https://www.youtube.com/watch?v=4rhShOPHl50", "canonicalUrl": "https://www.youtube.com/watch?v=4rhShOPHl50", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-20", "retrievedAt": "2026-09-29T13:38:07.599Z", "lastVerifiedAt": "2026-10-01T08:01:58.271Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-013", "title": "Salesforce explicitly prices Help Agent on successful outcomes", "slug": "salesforce-help-agent-outcome-pricing", "eventDate": "2026-09-04", "publicationDate": "2026-09-04", "discoveredDate": "2026-09-04", "datePrecision": "OBSERVED_BY", "companies": [ "Salesforce" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Salesforce documentation states that Help Agent pricing is strictly outcome-based. Unresolved sessions are not billed, while successful resolutions are billable outcomes.", "observedFact": "Salesforce documentation states that Help Agent pricing is strictly outcome-based. Unresolved sessions are not billed, while successful resolutions are billable outcomes.", "interpretation": "Direct commercial evidence for Law 2, because the charge is tied to the result produced instead of seat access or raw activity volume.", "mechanisms": [], "laws": [ "LAW_2", "LAW_3" ], "propositions": [ "COMMERCIAL_ALIGNMENT" ], "themes": [ "outcome-pricing", "verification" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "predictionTested": "As autonomous systems perform work, value capture can migrate from access and activity units toward successful outcomes.", "whyItMatters": "The outcome itself becomes the billing unit, while a verification rule determines whether payment is triggered.", "counterInterpretation": "Outcome pricing may remain limited to narrowly measurable customer-service use cases.", "whatWouldStrengthen": "Expansion of outcome pricing into more complex enterprise processes and a growing share of Salesforce monetisation.", "whatWouldWeaken": "If outcome pricing remains a marginal add-on while the economics continue to be overwhelmingly seat-based.", "relatedEvidenceIds": [ "AE-2026-003", "AE-2026-010", "AE-2026-011", "AE-2026-012" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.774Z", "sources": [ { "id": "dbd8b403-e0f5-4138-8484-89a524801c43", "evidenceEventId": "AE-2026-013", "sourceName": "Salesforce", "sourceUrl": "https://help.salesforce.com/s/articleView?id=ai.usage_flex_credits.htm&language=en_US&type=5", "canonicalUrl": "https://help.salesforce.com/s/articleView?id=ai.usage_flex_credits.htm&language=en_US&type=5", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-04", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:54.060Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-017", "title": "Atlassian prices customer-service AI at $1 per successful resolution", "slug": "atlassian-one-dollar-successful-resolution-pricing", "eventDate": "2026-09-04", "publicationDate": "2026-09-04", "discoveredDate": "2026-09-04", "datePrecision": "OBSERVED_BY", "companies": [ "Atlassian" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Atlassian states that Customer Service Management AI agent resolutions will be charged at $1 per successful resolution from 3 December 2026, only when the AI fully resolves a customer request without handing it to a human.", "observedFact": "Atlassian states that Customer Service Management AI agent resolutions will be charged at $1 per successful resolution from 3 December 2026, only when the AI fully resolves a customer request without handing it to a human. Resolution is judged by the system’s LLM judge.", "interpretation": "Strong direct evidence of outcome-based commercial redesign and of verification becoming part of the billing architecture.", "mechanisms": [], "laws": [ "LAW_2", "LAW_3" ], "propositions": [ "COMMERCIAL_ALIGNMENT" ], "themes": [ "outcome-pricing", "verification" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "predictionTested": "Autonomous software can move value capture from user access toward successful outcomes, with verification determining when the economic event has occurred.", "whyItMatters": "A second major SaaS vendor is independently adopting successful-resolution pricing, making the pattern less idiosyncratic.", "counterInterpretation": "The model is announced but billing has not begun, and customer service is unusually easy to measure.", "whatWouldStrengthen": "Meaningful paid resolution volume and expansion of outcome pricing into other workflows.", "whatWouldWeaken": "Low adoption or abandonment of the model after billing begins.", "relatedEvidenceIds": [], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.974Z", "sources": [ { "id": "38820f66-06ca-423b-b42e-89ce8c695908", "evidenceEventId": "AE-2026-017", "sourceName": "Atlassian", "headlineAtVerification": "Manage usage for AI agent resolutions | Customer Service Management | Atlassian Support", "sourceUrl": "https://support.atlassian.com/customer-service-management/docs/manage-usage-for-ai-agent-resolutions/", "canonicalUrl": "https://support.atlassian.com/customer-service-management/docs/manage-usage-for-ai-agent-resolutions/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-04", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:54.532Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-021", "title": "Blackstone private-credit fund redemptions highlight software/AI durability concerns", "slug": "blackstone-private-credit-redemptions-software-ai", "eventDate": "2026-09-03", "publicationDate": "2026-09-03", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Blackstone" ], "sectors": [ "private-credit", "software", "private-equity" ], "sectorLabels": [ "private-credit", "software", "private-equity" ], "summary": "The FT reported continued high redemption requests at Blackstone’s large retail private-credit fund and linked investor concern partly to exposure to leveraged software companies facing uncertainty over AI’s impact, while also identifying broader private-credit pressures.", "observedFact": "The FT reported continued high redemption requests at Blackstone’s large retail private-credit fund and linked investor concern partly to exposure to leveraged software companies facing uncertainty over AI’s impact, while also identifying broader private-credit pressures.", "interpretation": "Potential evidence that autonomy-related software durability concerns are propagating from operating-company valuations and loans into portfolio-level investor behaviour.", "mechanisms": [ "CSD", "HBI" ], "laws": [], "propositions": [ "REVENUE_DURABILITY" ], "themes": [ "private-credit-transmission" ], "classification": "AMBIGUOUS", "evidenceStrength": 3, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "Autonomy exposure can influence valuations and capital structures before operating revenue fully resets, including through credit and portfolio channels.", "whyItMatters": "It may represent a second-order capital-market consequence of software durability uncertainty.", "counterInterpretation": "Redemptions have multiple causes, including liquidity preferences, fund structure, leverage and general private-credit sentiment; AI cannot be isolated as the cause.", "whatWouldStrengthen": "Repeated evidence that software/AI exposure predicts fund flows, write-downs or credit performance relative to otherwise comparable assets.", "whatWouldWeaken": "If redemption and credit stress prove unrelated to software autonomy exposure.", "relatedEvidenceIds": [ "AE-2026-019", "AE-2026-035" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-09-04T23:07:58.720Z", "sources": [ { "id": "aa5f380b-ee8e-451e-b7e9-d2ec1baf0326", "evidenceEventId": "AE-2026-021", "sourceName": "Financial Times", "headlineAtVerification": "Blackstone stands firm with cap on private credit outflows", "sourceUrl": "https://www.ft.com/content/41017d78-2d1c-4a41-9dee-c9237179d616", "canonicalUrl": "https://www.ft.com/content/41017d78-2d1c-4a41-9dee-c9237179d616", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 2, "publicationDate": "2026-09-03", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-09-04T23:07:58.720Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": false, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-022", "title": "Big Law hours rise 4.2% despite widespread AI use, as clients press to leave the billable hour", "slug": "big-law-hours-rise-despite-ai-adoption", "eventDate": "2026-09-03", "publicationDate": "2026-09-03", "discoveredDate": "2026-09-05", "datePrecision": "EXACT", "companies": [ "Goldman Sachs", "Morgan Stanley", "Citigroup" ], "sectors": [ "legal", "financial-services" ], "sectorLabels": [ "legal", "financial-services" ], "summary": "Demand for law firm hours rose 4.2% through the first half of 2026 against a normal rate of about 1.5%, while two-thirds of firms reported daily associate use of AI and clients pressed to move away from the billable hour.", "observedFact": "Bloomberg Law reported that demand for law firm hours rose 4.2% through the first half of 2026, against normal demand growth of around 1.5%, according to Citi's law firm banking group. In a Citi survey of 57 firms, about three-quarters expect the billable hour model to be affected by AI by 2028 and about half said it already has been; two-thirds said associates use AI daily and more than half said partners do. Clients including private equity firms and Wall Street banks are pressing firms to reflect AI savings in fees. The column notes that if AI had already reduced the time required to service major clients, billed hours would be expected to fall instead of rising, and offers scope expansion, firms using freed time to do more or more thorough work, as one possible explanation.", "interpretation": "The observation is genuinely ambiguous for HBP and Revenue Durability. Buyer pressure on the billable hour is visible, while aggregate law-firm hours also rose. The latter does not show that human progression per matter failed to fall because demand and scope can increase at the same time. The record is therefore an early mixed signal to track. It does not show that the paper required progression to fall before commercial pressure could appear.", "mechanisms": [ "HBP" ], "laws": [ "LAW_2" ], "propositions": [ "REVENUE_DURABILITY", "COMMERCIAL_ALIGNMENT" ], "themes": [ "billable-hour", "scope-expansion", "outcome-pricing" ], "classification": "AMBIGUOUS", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "MISALIGNED", "signalStage": "EARLY_SIGNAL", "horizonNote": "Useful current evidence, but the legal-sector commercial-unit durability test remains longitudinal. Aggregate billed hours can rise even if effort per matter falls, so at this stage the observation cannot establish whether HBP-dependent pricing is durable or weakening.", "predictionTested": "Tests whether realised changes in human progression are visible alongside pressure on effort-based pricing, while recognising that aggregate hours are not the same measure as effort per unit of work.", "whyItMatters": "It is the clearest counterweight on the register to the legal pricing-pressure records. Billed hours rising at nearly three times the normal rate, while AI use is near-universal inside the same firms, is not what a simple reading of the mechanism predicts. It forces the sequence to be stated more carefully. Buyer pressure can run ahead of any measurable change in activity, driven by expectation and by the visible profitability of the supplier instead of an observed reduction in effort.", "counterInterpretation": "Hours could be rising because deal volume is rising, an M&A revival raises demand independently of how long each matter takes, so total hours may grow even while hours per unit of work fall. The measure reported is aggregate demand, which differs from effort per matter, and the framework's claim is about effort per matter. On that reading the record says little about progression and the mechanism remains untested here.", "whatWouldStrengthen": "Evidence that hours per comparable matter are flat or rising over several periods once deal volume is controlled for, or firms stating that AI-freed capacity is being redirected into expanded scope instead of lower fees.", "whatWouldWeaken": "Realised declines in hours billed per comparable matter, or a fall in aggregate hours once the deal cycle normalises, which would show the progression reduction arriving on the expected schedule and simply being masked by volume.", "falsificationRelevance": "Bears directly on the HBP falsification condition. If human progression in legal does not fall materially over the next several reporting periods while AI use remains near-universal, the mechanism's applicability to legal production is in question, and a simple delay would no longer explain it.", "relatedEvidenceIds": [ "AE-2026-008", "AE-2026-018", "AE-2026-041" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-05T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:49.180Z", "sources": [ { "id": "3bdef2b9-8bc1-46a2-8d3f-8e05a0e64acd", "evidenceEventId": "AE-2026-022", "sourceName": "Bloomberg Law", "headlineAtVerification": "Big Law AI Cost Savings Remain a Mystery as Billable Hours Jump", "sourceUrl": "https://news.bloomberglaw.com/business-and-practice/big-law-ai-cost-savings-remain-a-mystery-as-billable-hours-jump", "canonicalUrl": "https://news.bloomberglaw.com/business-and-practice/big-law-ai-cost-savings-remain-a-mystery-as-billable-hours-jump", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 2, "publicationDate": "2026-09-03", "retrievedAt": "2026-09-05T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:57.538Z", "verificationStatus": "VERIFIED_INDEPENDENT", "isPrimarySource": false, "supportsObservedFact": true } ] }, { "id": "AE-2026-020", "title": "FCA finds enterprise value depends increasingly on the AI harness around the model", "slug": "fca-frontier-ai-harness-governance", "eventDate": "2026-09-02", "publicationDate": "2026-09-02", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Financial Conduct Authority" ], "sectors": [ "financial-services", "software" ], "sectorLabels": [ "financial-services", "software" ], "summary": "The FCA published a multi-firm review focused on frontier AI, cyber resilience, harness engineering, governance and vulnerability management, highlighting that firms must structure controls and operating environments around frontier models.", "observedFact": "The FCA published a multi-firm review focused on frontier AI, cyber resilience, harness engineering, governance and vulnerability management, highlighting that firms must structure controls and operating environments around frontier models.", "interpretation": "Strong independent support for Law 1 and Law 3: the surrounding harness, controls, tooling and human oversight become central to making increasingly abundant model capability usable.", "mechanisms": [ "CSD" ], "laws": [ "LAW_1", "LAW_3" ], "propositions": [], "themes": [ "orchestration", "governance", "verification", "absorption" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "As model capability diffuses, value and operational scarcity should move toward the systems that organise, constrain, validate and absorb that capability.", "whyItMatters": "A regulator is examining the harness and control environment as a distinct operational layer, and does not treat the model alone as the relevant system.", "counterInterpretation": "Regulatory emphasis on controls does not necessarily show where profit or commercial value ultimately accumulates.", "whatWouldStrengthen": "Evidence that harness, governance and verification layers attract disproportionate enterprise spend, pricing power or strategic differentiation.", "whatWouldWeaken": "If model capability itself remains the dominant source of economic differentiation and surrounding controls commoditise.", "relatedEvidenceIds": [ "AE-2026-005", "AE-2026-006", "AE-2026-007", "AE-2026-015" ], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:49.110Z", "sources": [ { "id": "e32fee55-ace0-40f9-976f-9eff7f9ea6e8", "evidenceEventId": "AE-2026-020", "sourceName": "Financial Conduct Authority", "headlineAtVerification": "Frontier AI and cyber resilience | FCA", "sourceUrl": "https://www.fca.org.uk/publications/multi-firm-reviews/frontier-ai-cyber-resilience", "canonicalUrl": "https://www.fca.org.uk/publications/multi-firm-reviews/frontier-ai-cyber-resilience", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-02", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:56.458Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-018", "title": "Wall Street banks push Big Law to pass AI productivity into lower fees", "slug": "wall-street-banks-big-law-ai-fees", "eventDate": "2026-09-01", "publicationDate": "2026-09-01", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Goldman Sachs", "Morgan Stanley", "Citigroup" ], "sectors": [ "legal", "financial-services", "consulting" ], "sectorLabels": [ "legal", "financial-services", "consulting" ], "summary": "The FT reported that Morgan Stanley, Citigroup and Goldman Sachs are pressing major law firms to reflect AI-enabled efficiencies in fees, including through competitive bidding, fixed fees and requests for evidence of AI-driven savings.", "observedFact": "The FT reported that Morgan Stanley, Citigroup and Goldman Sachs are pressing major law firms to reflect AI-enabled efficiencies in fees, including through competitive bidding, fixed fees and requests for evidence of AI-driven savings.", "interpretation": "Very strong evidence that reduced human progression is translating into buyer pressure on the historic billable-hour commercial unit.", "mechanisms": [ "HBP" ], "laws": [ "LAW_2", "LAW_3" ], "propositions": [ "AGENTIC_PROFIT_PARADOX", "REVENUE_DURABILITY" ], "themes": [ "outcome-pricing" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "commercialAlignment": "MISALIGNED", "predictionTested": "Commercial-unit pressure should appear once buyers can observe that less human progression is required, even if demand for high-value legal judgement remains.", "whyItMatters": "This is a direct buyer response to productivity change that is already happening, with no reliance on forecasts about future legal pricing.", "counterInterpretation": "Large financial institutions may have exceptional bargaining power and may not represent the broader legal market.", "whatWouldStrengthen": "Observed reductions in realised matter fees, broader alternative-fee adoption and sustained hourly-share decline.", "whatWouldWeaken": "If hourly billing remains dominant and economically durable despite substantial realised AI productivity.", "relatedEvidenceIds": [ "AE-2026-008", "AE-2026-022", "AE-2026-041" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:49.041Z", "sources": [ { "id": "94722489-4bb9-421b-9558-a88de15f6d9b", "evidenceEventId": "AE-2026-018", "sourceName": "Financial Times", "headlineAtVerification": "Wall Street banks push Big Law to cut fees because of AI", "sourceUrl": "https://www.ft.com/content/5240a6ac-b2e8-4897-a0a4-cbc7fc283bc9", "canonicalUrl": "https://www.ft.com/content/5240a6ac-b2e8-4897-a0a4-cbc7fc283bc9", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 2, "publicationDate": "2026-09-01", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-09-04T23:07:58.720Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": false, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-019", "title": "FT reports PE-backed software paying more to extend debt amid AI durability concerns", "slug": "pe-backed-software-debt-extension-ai-durability", "eventDate": "2026-09-01", "publicationDate": "2026-09-01", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Proofpoint", "Sophos", "Thoma Bravo" ], "sectors": [ "software", "private-equity", "private-credit" ], "sectorLabels": [ "software", "private-equity", "private-credit" ], "summary": "The FT reported that PE-backed software companies are using shorter amend-and-extend transactions, paying higher yields and accepting stronger creditor protections as lenders assess AI-related threats to long-term software business-model durability alongside leverage and maturity risks.", "observedFact": "The FT reported that PE-backed software companies are using shorter amend-and-extend transactions, paying higher yields and accepting stronger creditor protections as lenders assess AI-related threats to long-term software business-model durability alongside leverage and maturity risks.", "interpretation": "Evidence that perceived Revenue Durability under autonomy can affect financing conditions before aggregate revenue collapse.", "mechanisms": [ "CSD", "HBI" ], "laws": [], "propositions": [ "REVENUE_DURABILITY" ], "themes": [ "capital-markets-transmission" ], "classification": "AMBIGUOUS", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "Markets and creditors can reprice the durability of exposed commercial structures before full operating-model or revenue deterioration is visible.", "whyItMatters": "It suggests a transmission channel from autonomy exposure into cost of capital and creditor protections.", "counterInterpretation": "Leverage, maturity walls, interest rates and CLO dynamics may explain much of the financing pressure independently of AI.", "whatWouldStrengthen": "Credit spreads, covenants or refinancing outcomes that systematically differ by measurable autonomy exposure after controlling for leverage and sector fundamentals.", "whatWouldWeaken": "If AI concerns disappear from underwriting while financing stress persists for conventional credit reasons.", "relatedEvidenceIds": [ "AE-2026-021", "AE-2026-035" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-09-04T23:07:58.720Z", "sources": [ { "id": "d16d6731-5f7f-46bc-85fe-b12aaa1e11e8", "evidenceEventId": "AE-2026-019", "sourceName": "Financial Times", "sourceUrl": "https://www.ft.com/content/485a3d1c-d282-4921-9dee-c644c5856030", "canonicalUrl": "https://www.ft.com/content/485a3d1c-d282-4921-9dee-c644c5856030", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 2, "publicationDate": "2026-09-01", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-09-04T23:07:58.720Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": false, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-014", "title": "FT reports consulting clients cutting external work and pressing fees as AI expands internal capability", "slug": "ft-consulting-clients-insourcing-fee-pressure", "eventDate": "2026-08-31", "publicationDate": "2026-08-31", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "KPMG", "Accenture", "Deloitte", "EY", "PwC", "Bayer", "Bristol Myers Squibb", "Commerzbank" ], "sectors": [ "consulting" ], "sectorLabels": [ "consulting" ], "summary": "The FT reported that companies are bringing more technology work in-house, reducing reliance on large consulting teams and demanding lower, fixed or performance-linked fees as AI reduces the human effort required for implementation and analysis.", "observedFact": "The FT reported that companies are bringing more technology work in-house, reducing reliance on large consulting teams and demanding lower, fixed or performance-linked fees as AI reduces the human effort required for implementation and analysis.", "interpretation": "Direct evidence of the Agentic Profit Paradox mechanism in consulting. Autonomous capability raises productivity while weakening the commercial defensibility of human-progression-based delivery units.", "mechanisms": [ "HBP", "CSD" ], "laws": [ "LAW_2", "LAW_3" ], "propositions": [ "AGENTIC_PROFIT_PARADOX", "REVENUE_DURABILITY" ], "themes": [ "outcome-pricing" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "commercialAlignment": "MISALIGNED", "predictionTested": "Consulting can remain in demand while billable progression becomes less defensible and value moves toward senior judgement, leadership advice and measurable outcomes.", "whyItMatters": "This is buyer behaviour, which carries more weight than a vendor prediction. Clients are connecting AI-enabled productivity directly to lower consultant dependence and pricing pressure.", "counterInterpretation": "Consulting pressure may also reflect procurement cycles, macro conditions and broader cost-cutting.", "whatWouldStrengthen": "Realised fee compression, smaller delivery teams, greater insourcing and higher outcome-based revenue share.", "whatWouldWeaken": "If AI materially reduces delivery labour but consulting pricing and margins remain structurally unchanged.", "relatedEvidenceIds": [], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.840Z", "sources": [ { "id": "f8fe2f8f-f6d6-4a04-9112-9dd6dbd2d4d4", "evidenceEventId": "AE-2026-014", "sourceName": "Financial Times", "sourceUrl": "https://www.ft.com/content/709d7e2d-0c2f-45bc-b559-e7328372cf9e", "canonicalUrl": "https://www.ft.com/content/709d7e2d-0c2f-45bc-b559-e7328372cf9e", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 2, "publicationDate": "2026-08-31", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-09-04T23:07:58.720Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": false, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-015", "title": "FSB elevates autonomous-AI control and resilience into G20 financial-system discussion", "slug": "fsb-autonomous-ai-g20-resilience", "eventDate": "2026-08-31", "publicationDate": "2026-08-31", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Financial Stability Board" ], "sectors": [ "financial-services" ], "sectorLabels": [ "financial-services" ], "summary": "The FSB Chair warned G20 finance ministers and central bank governors that increasingly autonomous frontier AI may materially change the speed, scale and economics of cyber risk and stressed resilience and responsible deployment.", "observedFact": "The FSB Chair warned G20 finance ministers and central bank governors that increasingly autonomous frontier AI may materially change the speed, scale and economics of cyber risk and stressed resilience and responsible deployment.", "interpretation": "Independent support for Law 3: as autonomous capability scales, control, resilience, authorisation and accountability become more economically consequential.", "mechanisms": [], "laws": [ "LAW_3" ], "propositions": [], "themes": [ "governance", "verification", "judgement", "boundary-conditions" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "The layer that constrains and governs autonomous throughput should become more important as the scale and consequences of autonomous action increase.", "whyItMatters": "Governance is moving from enterprise policy into systemic-risk architecture.", "counterInterpretation": "Systemic-risk importance does not automatically imply that governance captures commercial value.", "whatWouldStrengthen": "Evidence of sustained spending, pricing power and strategic differentiation in governance/verification layers.", "whatWouldWeaken": "If governance becomes standardised and low-cost without becoming a meaningful economic layer.", "relatedEvidenceIds": [ "AE-2026-005", "AE-2026-006", "AE-2026-007", "AE-2026-020" ], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-09-04T23:07:58.720Z", "sources": [ { "id": "1dcd258a-eb55-4360-99cc-7c9156af7444", "evidenceEventId": "AE-2026-015", "sourceName": "Financial Stability Board", "headlineAtVerification": "FSB Chair’s letter to G20 Finance Ministers and Central Bank Governors: August 2026 - Financial Stability Board", "sourceUrl": "https://www.fsb.org/2026/08/fsb-chairs-letter-to-g20-finance-ministers-and-central-bank-governors-august-2026/", "canonicalUrl": "https://www.fsb.org/2026/08/fsb-chairs-letter-to-g20-finance-ministers-and-central-bank-governors-august-2026/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-08-31", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:54.312Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-016", "title": "Atlassian introduces metered Rovo credits for AI and Teamwork Graph usage", "slug": "atlassian-rovo-credits-usage-pricing", "eventDate": "2026-08-31", "publicationDate": "2026-09-04", "discoveredDate": "2026-09-04", "datePrecision": "EFFECTIVE_PRICING_TABLE", "companies": [ "Atlassian" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Atlassian documents Rovo credits as a meter for AI usage and enriched Teamwork Graph context, including calls made through the Rovo MCP server and CLI.", "observedFact": "Atlassian documents Rovo credits as a meter for AI usage and enriched Teamwork Graph context, including calls made through the Rovo MCP server and CLI. The pricing table applies to usage and forecasts from 31 August 2026, with extra-usage billing beginning 3 December 2026.", "interpretation": "Evidence that SaaS monetisation is extending beyond seats toward autonomous activity, context and agent-mediated access to underlying platform capability.", "mechanisms": [ "HBI", "CSD" ], "laws": [ "LAW_1", "LAW_2" ], "propositions": [], "themes": [ "usage-pricing", "orchestration" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "predictionTested": "As agents, in place of users, consume software capability, the seat becomes a weaker standalone measure of value and new activity and context units emerge.", "whyItMatters": "Atlassian is monetising agent and third-party access to underlying context even when work takes place outside the conventional product interface.", "counterInterpretation": "Usage pricing may simply supplement seats, leaving them neither replaced nor weakened.", "whatWouldStrengthen": "A rising share of Atlassian revenue tied to agent/context usage and lower relative importance of user-seat expansion.", "whatWouldWeaken": "If usage meters remain minor while seat economics remain unchanged.", "relatedEvidenceIds": [], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.909Z", "sources": [ { "id": "ee108da9-6105-46ea-9052-565cef6e7234", "evidenceEventId": "AE-2026-016", "sourceName": "Atlassian", "headlineAtVerification": "How Rovo credits work | Rovo | Atlassian Support", "sourceUrl": "https://support.atlassian.com/rovo/docs/rovo-usage-limits/", "canonicalUrl": "https://support.atlassian.com/rovo/docs/rovo-usage-limits/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-04", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:54.427Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-044", "title": "Open-weight models take most Vercel gateway tokens while frontier providers retain most spend", "slug": "vercel-open-weight-volume-frontier-spend-concentration", "eventDate": "2026-08-31", "publicationDate": "2026-09-17", "discoveredDate": "2026-09-30", "datePrecision": "OBSERVED_BY", "companies": [ "Vercel" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Vercel's September 2026 AI Gateway Production Index shows open-weight models processing 56% of gateway token volume in August while accounting for 14% of estimated spend. Average price per token fell 23.2% during the month, while Anthropic retained 64% of gateway spend. The same dataset therefore shows both rapid diffusion of lower-cost capability and persistent concentration of premium spending.", "observedFact": "In August 2026, open-weight models processed 56% of token volume on Vercel AI Gateway and accounted for 14% of estimated spend. Vercel reports that average price per token fell 23.2% in August. Anthropic nevertheless retained 64% of all estimated gateway spend, and its models had held the top two positions by spend each month since December.", "interpretation": "This is genuinely ambiguous evidence for Capability Scarcity Dependence. Open-weight models taking a majority of production token volume while average inference cost falls is consistent with capability becoming more available and cheaper. At the same time, the persistence of concentrated frontier spending shows that scarcity rents can survive at the high-value end even while lower-cost capability diffuses rapidly. The observation therefore argues against treating commoditisation as uniform.", "mechanisms": [ "CSD" ], "laws": [], "propositions": [], "themes": [ "open-weight-models", "frontier-premium", "model-routing", "inference-cost" ], "classification": "AMBIGUOUS", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "horizonNote": "This is an early mixed CSD signal. The paper explicitly allows capability scarcity to weaken unevenly, so open-weight volume and persistent frontier spend should be tracked over time instead of being treated as a new boundary to the theory.", "predictionTested": "Tests whether broad availability of cheaper and open-weight capability erodes scarcity rents, or whether premium frontier capability continues to sustain concentrated spending.", "whyItMatters": "The same production dataset contains evidence in both directions. It is therefore useful for defining where CSD may weaken first and where premium capability can remain scarce enough to command materially higher spending.", "counterInterpretation": "Vercel AI Gateway is one routing platform and does not represent the whole enterprise AI market. Estimated spend uses public list prices in place of actual negotiated bills, and differences in workload complexity can justify premium model usage without establishing durable scarcity rents.", "whatWouldStrengthen": "Comparable production-routing data across several gateways or enterprise datasets showing whether frontier spend concentration persists as open-weight capability improves.", "whatWouldWeaken": "A sustained collapse in frontier spend share, rapid multi-provider switching without performance loss, or evidence that premium model prices converge towards open-weight inference costs.", "falsificationRelevance": "If cheap substitute capability becomes widely available while premium providers retain persistent pricing power and spend concentration across comparable workloads, the CSD proposition would need to be narrowed.", "relatedEvidenceIds": [], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-30T17:06:51.127Z", "createdAt": "2026-09-30T17:06:51.127Z", "updatedAt": "2026-10-11T09:24:49.742Z", "sources": [ { "id": "602a586d-141f-41ac-b41e-e550dc01ba1c", "evidenceEventId": "AE-2026-044", "sourceName": "Vercel", "headlineAtVerification": "Open-weight models take 56% of token volume, Astra doubles Fable 5.1 spend", "sourceUrl": "https://vercel.com/blog/ai-gateway-production-index-september-2026", "canonicalUrl": "https://vercel.com/blog/ai-gateway-production-index-september-2026", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-17", "retrievedAt": "2026-09-30T17:06:51.127Z", "lastVerifiedAt": "2026-10-01T08:01:58.451Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-011", "title": "Salesforce and Anthropic announce Claudeforce", "slug": "salesforce-anthropic-claudeforce", "eventDate": "2026-08-26", "publicationDate": "2026-08-26", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Anthropic", "Salesforce" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Salesforce and Anthropic announced Claudeforce, making Salesforce data, workflows, business logic, actions and governance accessible directly in Claude, initially through a plugin with 37 prebuilt sales skills.", "observedFact": "Salesforce and Anthropic announced Claudeforce, making Salesforce data, workflows, business logic, actions and governance accessible directly in Claude, initially through a plugin with 37 prebuilt sales skills.", "interpretation": "Further evidence that value can migrate away from the proprietary application interface while the underlying data, workflow, governance and action layers remain central.", "mechanisms": [ "HBI" ], "laws": [ "LAW_1", "LAW_3" ], "propositions": [], "themes": [ "interface-disintermediation", "orchestration", "governance" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "predictionTested": "External agents can become the user-facing interaction layer while the incumbent protects economic relevance by becoming the governed execution and context layer underneath.", "whyItMatters": "The incumbent is voluntarily allowing a third-party agent to become the interface, giving up the attempt to preserve UI scarcity.", "counterInterpretation": "The partnership may increase Salesforce engagement without materially changing the role of seats or interfaces.", "whatWouldStrengthen": "Evidence that users increasingly complete Salesforce work from Claude while Salesforce maintains or improves monetisation.", "whatWouldWeaken": "If Claudeforce remains a niche integration with little behavioural or commercial impact.", "relatedEvidenceIds": [ "AE-2026-003", "AE-2026-010", "AE-2026-012", "AE-2026-013" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.706Z", "sources": [ { "id": "4a5ba747-97a8-4d3a-906d-7c470e0c0f9d", "evidenceEventId": "AE-2026-011", "sourceName": "Salesforce", "headlineAtVerification": "Salesforce and Anthropic Announce Claudeforce - Salesforce", "sourceUrl": "https://www.salesforce.com/uk/news/press-releases/2026/08/26/salesforce-and-anthropic-announce-claudeforce/?bc=OTH", "canonicalUrl": "https://www.salesforce.com/uk/news/press-releases/2026/08/26/salesforce-and-anthropic-announce-claudeforce/?bc=OTH", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-08-26", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:53.407Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-012", "title": "Salesforce Q2 results show strong durability during commercial and architectural redesign", "slug": "salesforce-q2-fy27-durability-during-redesign", "eventDate": "2026-08-26", "publicationDate": "2026-08-26", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Salesforce" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Salesforce reported $10.8 billion in subscription and support revenue, up 12% year on year, nearly $3.9 billion in Agentforce and Data 360 ARR, 7 billion Agentic Work Units delivered to date, and reorganised revenue disclosure around Agentforce Apps and Data 360, Headless Platform and Other.", "observedFact": "Salesforce reported $10.8 billion in subscription and support revenue, up 12% year on year, nearly $3.9 billion in Agentforce and Data 360 ARR, 7 billion Agentic Work Units delivered to date, and reorganised revenue disclosure around Agentforce Apps and Data 360, Headless Platform and Other.", "interpretation": "Strong comparative evidence that redesign can coexist with revenue durability. The result is not a contradiction to the framework because Salesforce is changing both activity architecture and value-capture architecture.", "mechanisms": [ "HBI" ], "laws": [ "LAW_1", "LAW_2" ], "propositions": [ "REVENUE_DURABILITY", "COMMERCIAL_ALIGNMENT" ], "themes": [], "classification": "BOUNDARY", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "Incumbent strength during redesign is consistent with the paper's distinction between deployment strength and business-model durability. The observation is tracked comparatively and does not count as weakening merely because the incumbent remains strong.", "predictionTested": "Firms that realign commercial structure with autonomous activity can remain economically durable while interface dependence and conventional activity units change.", "whyItMatters": "It is a live counterexample to a simplistic 'agentic AI destroys SaaS' thesis and therefore a useful positive test of the redesign proposition.", "counterInterpretation": "Growth may be explained by acquisition effects, installed-base strength and conventional subscription expansion.", "whatWouldStrengthen": "Sustained growth in autonomous activity and non-seat monetisation alongside stable retention and margins.", "whatWouldWeaken": "If redesign proves commercially immaterial or fails to support durable growth.", "relatedEvidenceIds": [ "AE-2026-003", "AE-2026-010", "AE-2026-011", "AE-2026-013" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-09-04T23:07:58.720Z", "sources": [ { "id": "cd3b71b7-082e-48e7-9537-8d6dc1088e4c", "evidenceEventId": "AE-2026-012", "sourceName": "Salesforce", "headlineAtVerification": "Salesforce Delivers Record Second Quarter Fiscal 2027 Results - Salesforce", "sourceUrl": "https://www.salesforce.com/news/press-releases/2026/08/26/fy27-q2-earnings/?bc=OTH", "canonicalUrl": "https://www.salesforce.com/news/press-releases/2026/08/26/fy27-q2-earnings/?bc=OTH", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-08-26", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:53.700Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-010", "title": "Salesforce expands Headless 360 so agents can invoke enterprise capabilities directly", "slug": "salesforce-headless-360-agent-access", "eventDate": "2026-08-25", "publicationDate": "2026-08-25", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Salesforce" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Salesforce expanded Headless 360 across its platform, exposing reusable enterprise capabilities through MCP, developer tools and headless experiences so authorised AI agents can discover and take action without relying on conventional application interfaces.", "observedFact": "Salesforce expanded Headless 360 across its platform, exposing reusable enterprise capabilities through MCP, developer tools and headless experiences so authorised AI agents can discover and take action without relying on conventional application interfaces.", "interpretation": "Direct evidence of HBI compression and of an incumbent repositioning value from the screen toward data, workflow, business logic, permissions and execution.", "mechanisms": [ "HBI" ], "laws": [ "LAW_1" ], "propositions": [ "COMMERCIAL_ALIGNMENT" ], "themes": [ "interface-disintermediation", "orchestration" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "predictionTested": "As agents become an interaction layer, interface dependence weakens and durable value can migrate toward underlying systems of record, workflow and governed action.", "whyItMatters": "The vendor is explicitly designing for software to remain valuable even when humans do not navigate its interface directly.", "counterInterpretation": "Headless access may complement conventional user interfaces instead of displacing them.", "whatWouldStrengthen": "Rising agent-mediated activity, lower relative UI usage and sustained monetisation of underlying data/actions.", "whatWouldWeaken": "If headless channels remain marginal and value continues to be captured mainly through human-facing seats.", "relatedEvidenceIds": [ "AE-2026-003", "AE-2026-011", "AE-2026-012", "AE-2026-013" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.641Z", "sources": [ { "id": "f60897fc-4a8c-489f-b60c-c77dcee59a43", "evidenceEventId": "AE-2026-010", "sourceName": "Salesforce", "headlineAtVerification": "Salesforce Turns Enterprise Applications into Enterprise Capabilities - Salesforce", "sourceUrl": "https://www.salesforce.com/ap/news/press-releases/2026/08/25/salesforce-turns-enterprise-applications-into-enterprise-capabilities/?bc=OTH", "canonicalUrl": "https://www.salesforce.com/ap/news/press-releases/2026/08/25/salesforce-turns-enterprise-applications-into-enterprise-capabilities/?bc=OTH", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-08-25", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:53.110Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-042", "title": "Futurum survey finds buyers still prefer per-seat pricing for separately billed generative AI", "slug": "futurum-buyers-prefer-seat-pricing-for-ai-addons", "eventDate": "2026-08-17", "publicationDate": "2026-08-17", "discoveredDate": "2026-09-30", "datePrecision": "EXACT", "companies": [], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Futurum's 2H 2026 Enterprise Software Decision-Makers Survey finds a split in buyer pricing preferences. Per-seat pricing ranks last for core software at 12.6%, but first for separately billed generative AI functionality at 42.3%, ahead of consumption pricing at 36.6% and outcome pricing at 21.1%.", "observedFact": "Among buyers surveyed by Futurum, per-seat pricing was selected by 12.6% for core software. Among buyers whose vendors charge separately for generative AI functionality, per-seat pricing ranked first at 42.3%, ahead of consumption at 36.6% and outcomes at 21.1%.", "interpretation": "This is a meaningful boundary condition for Law 2. Buyers can reject seat pricing for established core software while preferring the predictability of a seat-based line item for a newer AI cost category. It suggests that budget certainty and cost volatility can preserve access-based pricing even as autonomous capability expands.", "mechanisms": [], "laws": [ "LAW_2" ], "propositions": [], "themes": [ "software-pricing", "buyer-preference", "seat-pricing", "cost-predictability" ], "classification": "BOUNDARY", "evidenceStrength": 3, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "Buyer preference for predictable per-seat AI pricing is a current persistence signal. The paper allows access-based models to persist, so the longer-term question is whether access remains as defensible a basis for value as autonomous activity becomes commercially meaningful.", "predictionTested": "Tests whether buyers move consistently away from access-based pricing as autonomous capability expands, or whether uncertainty can preserve seat pricing in some categories.", "whyItMatters": "The same buyers express different commercial preferences for core software and AI. That makes the observation more useful than a simple vendor pricing example, because it shows that the commercial unit can depend on the risk characteristics of the category as well as on the underlying technical capability.", "counterInterpretation": "This is a stated-preference survey and does not record transactions. Non-seat models still account for 57.7% of AI-specific preferences combined, and the result may reflect temporary uncertainty over AI cost instead of durable access-based economics.", "whatWouldStrengthen": "Transaction-level evidence showing that per-seat AI pricing remains preferred and commercially durable as usage becomes predictable and agent volumes scale.", "whatWouldWeaken": "A sustained move towards consumption, action or outcome pricing once buyers gain better cost visibility, or evidence that per-seat preference does not translate into actual contracts.", "falsificationRelevance": "If per-seat pricing remains the dominant buyer preference and commercial model for autonomous AI over several years, it would weigh directly against a strong version of Law 2.", "relatedEvidenceIds": [], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-30T16:07:16.128Z", "createdAt": "2026-09-30T16:07:16.128Z", "updatedAt": "2026-10-11T09:24:49.605Z", "sources": [ { "id": "43b9bcc6-bfd6-4702-93a0-0aa100b7495f", "evidenceEventId": "AE-2026-042", "sourceName": "Futurum", "headlineAtVerification": "The Pricing Split: Why Core Software and AI Can't Share One Commercial Model", "sourceUrl": "https://futurumgroup.com/insights/the-pricing-split-why-core-software-and-ai-cant-share-one-commercial-model/", "canonicalUrl": "https://futurumgroup.com/insights/the-pricing-split-why-core-software-and-ai-cant-share-one-commercial-model/", "sourceType": "INSTITUTIONAL_RESEARCH", "sourceTier": 3, "publicationDate": "2026-08-17", "retrievedAt": "2026-09-30T16:07:16.128Z", "lastVerifiedAt": "2026-10-01T08:01:58.885Z", "verificationStatus": "VERIFIED_INSTITUTIONAL", "isPrimarySource": false, "supportsObservedFact": true } ] }, { "id": "AE-2026-009", "title": "DeepSeek open-sources modular agent harness", "slug": "deepseek-open-source-agent-harness", "eventDate": "2026-08-13", "publicationDate": "2026-08-13", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "DeepSeek" ], "sectors": [ "financial-services", "software" ], "sectorLabels": [ "financial-services", "software" ], "summary": "DeepSeek released an open-source agent harness in developer preview under the MIT licence, built around an everything-is-a-plugin architecture.", "observedFact": "DeepSeek released an open-source agent harness in developer preview under the MIT licence, built around an everything-is-a-plugin architecture.", "interpretation": "Evidence that capability diffusion is moving beyond model weights into reusable agent orchestration and harness infrastructure, weakening scarcity in previously specialised implementation layers.", "mechanisms": [ "CSD" ], "laws": [ "LAW_1" ], "propositions": [], "themes": [ "open-source", "orchestration" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "As autonomous capability diffuses through open systems, scarcity should move away from raw access to the capability and toward how it is integrated, governed and applied.", "whyItMatters": "Open agent infrastructure broadens access to execution capability as well as to model inference.", "counterInterpretation": "Developer-preview software may have little near-term enterprise economic impact.", "whatWouldStrengthen": "Meaningful enterprise adoption, ecosystem growth, substitution of proprietary orchestration layers and lower deployment costs.", "whatWouldWeaken": "If open harnesses remain technically interesting but economically marginal.", "relatedEvidenceIds": [], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.573Z", "sources": [ { "id": "faf5e640-374b-42a1-8933-785af2d56c31", "evidenceEventId": "AE-2026-009", "sourceName": "DeepSeek / GitHub", "headlineAtVerification": "GitHub - deepseek-ai/deepseek-harness: DeepSeek Harness: Everything is a Plugin. · GitHub", "sourceUrl": "https://github.com/deepseek-ai/deepseek-harness", "canonicalUrl": "https://github.com/deepseek-ai/deepseek-harness", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-08-13", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:52.805Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-024", "title": "The Economist reports that China's AI drive threatens the world's largest workforce", "slug": "china-s-ai-drive-threatens-the-world-s-largest-workforce-the-economist", "eventDate": "2026-08-06", "publicationDate": "2026-08-06", "discoveredDate": "2026-09-28", "datePrecision": "EXACT", "companies": [], "sectors": [], "sectorLabels": [], "summary": "ByteDance released Seedance 2.0 AI model in February 2026. Microdramas now take 'days to produce instead of weeks' and are 'up to 90% cheaper (the bill is mostly tokens, instead of humans).' Li Dazhi's daily salary 'fell by half' and 'gigs dried up.' He and colleagues switched to unrelated work (insurance sales, vendor stalls, food delivery, rideshare). The microdrama industry had 700,000 direct j", "observedFact": "ByteDance released Seedance 2.0 AI model in February 2026. Microdramas now take 'days to produce instead of weeks' and are 'up to 90% cheaper (the bill is mostly tokens, instead of humans).' Li Dazhi's daily salary 'fell by half' and 'gigs dried up.' He and colleagues switched to unrelated work (insurance sales, vendor stalls, food delivery, rideshare). The microdrama industry had 700,000 direct jobs and 1.3m indirect ones as of 2025; this was created before the AI capability entered.", "interpretation": "This demonstrates HBP, where human progression through stages of work is now replaced by autonomous capability. Video production that required weeks of human labour and human skill progression now requires days and is tokenised. It shows CSD (the scarcity of video production and editing expertise, which supported job creation and wage income, has declined). The economic consequence is measurable. Worker income halved, work unit cost fell 90%, and workers were displaced into lower-skill service work. This is a shift in the structure of how work is compensated and organized.", "mechanisms": [ "HBP", "CSD" ], "laws": [], "propositions": [ "AGENTIC_PROFIT_PARADOX" ], "themes": [], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "predictionTested": "Tests HBP and CSD, asking whether humans carrying work through stages (scriptwriting, shooting, editing and post-production) can be replaced by a single autonomous system, and whether the scarcity of human video production skill is declining. The text shows both. Production time collapsed, cost per unit fell 90%, and workers were displaced. Worker transition to lower-wage work implies that video production expertise no longer commands the same wage premium.", "whyItMatters": "This demonstrates HBP, where human progression through stages of work is now replaced by autonomous capability. Video production that required weeks of human labour and human skill progression now requires days and is tokenised. It shows CSD (the scarcity of video production and editing expertise, which supported job creation and wage income, has declined). The economic consequence is measurable. Worker income halved, work unit cost fell 90%, and workers were displaced into lower-skill service work. This is a shift in the structure of how work is compensated and organized.", "counterInterpretation": "The productivity and cost improvement are real, but the job displacement could reflect a temporary labour market shock instead of structural value movement. Workers may redeploy to video curation, AI training, or higher-value uses. The 90% cost reduction benefits consumers and producers of microdramas, and not necessarily a capital owner or orchestrator, so the value may not be concentrating.", "relatedEvidenceIds": [], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-28T13:05:26.912Z", "createdAt": "2026-09-28T13:04:46.193Z", "updatedAt": "2026-10-11T09:24:49.326Z", "sources": [ { "id": "e3fbac91-edcd-4b27-895e-c2ef63bc0017", "evidenceEventId": "AE-2026-024", "sourceName": "economist.com", "sourceUrl": "https://www.economist.com/briefing/2026/08/06/chinas-ai-drive-threatens-the-worlds-largest-workforce", "canonicalUrl": "https://www.economist.com/briefing/2026/08/06/chinas-ai-drive-threatens-the-worlds-largest-workforce", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 2, "publicationDate": "2026-08-06", "retrievedAt": "2026-09-28T13:04:46.622Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": false, "supportsObservedFact": false, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-023", "title": "EY survey finds C-suites pivoting from AI adoption to unlocking value as token costs trigger fiscal scrutiny", "slug": "ey-survey-c-suites-pivot-from-ai-adoption-to-unlocking-value-as-escala", "eventDate": "2026-07-28", "publicationDate": "2026-07-28", "discoveredDate": "2026-09-28", "datePrecision": "EXACT", "companies": [], "sectors": [], "sectorLabels": [], "summary": "98% of senior leaders using AI tools that require tokens say token usage costs caused their organisation to reconsider their approach. 82% are concerned about AI token usage and costs. Only 64% actively monitor token usage and have clear budgetary guardrails. 76% say off-the-shelf software solutions no longer meet their needs. 91% view in-house AI-built software as critical. 87% have fully deploye", "observedFact": "98% of senior leaders using AI tools that require tokens say token usage costs caused their organisation to reconsider their approach. 82% are concerned about AI token usage and costs. Only 64% actively monitor token usage and have clear budgetary guardrails. 76% say off-the-shelf software solutions no longer meet their needs. 91% view in-house AI-built software as critical. 87% have fully deployed or are piloting in-house AI-built software development. 82% anticipate traditional per-seat SaaS pricing model will become less relevant in their industry within five years. Actual AI spending fell short of projections. 35% of leaders projected spending 10 million dollars or more, and only 23% report that spend today. 18% planned 50% or more of total budget to AI, and only 3% have done so.", "interpretation": "Token consumption costs are driving enterprises away from vendor commercial software toward in-house autonomous systems, and consumption-based cost drivers are making traditional per-seat licensing incompatible with autonomous activity structures. Value capture is restructuring around cost governance of autonomous throughput, and less around access.", "mechanisms": [], "laws": [ "LAW_2" ], "propositions": [ "AGENTIC_PROFIT_PARADOX", "REVENUE_DURABILITY", "CAPITAL_STRUCTURED_AUTONOMY" ], "themes": [], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "predictionTested": "Enterprises actively building in-house autonomous systems in response to consumption cost structures incompatible with traditional SaaS licensing. Per-seat pricing model anticipated to lose relevance. Spending projections missing targets suggest commercial structures are not sustaining projected revenue.", "whyItMatters": "Token consumption costs are driving enterprises away from vendor commercial software toward in-house autonomous systems, and consumption-based cost drivers are making traditional per-seat licensing incompatible with autonomous activity structures. Value capture is restructuring around cost governance of autonomous throughput, and less around access.", "counterInterpretation": "The move to in-house development could reflect a general enterprise preference for customisation, with no structural change driven by AI. The gap between projected and actual spending could indicate that enterprises are risk-averse or cautious, without the commercial model failing. Token cost concerns may be temporary pain points in a new market, and may not point to durable repricing.", "relatedEvidenceIds": [], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-28T13:05:23.315Z", "createdAt": "2026-09-28T10:00:10.751Z", "updatedAt": "2026-10-11T09:24:49.260Z", "sources": [ { "id": "99635176-7988-4bd5-b8fa-c13ca6669245", "evidenceEventId": "AE-2026-023", "sourceName": "ey.com", "sourceUrl": "https://www.ey.com/en_us/newsroom/2026/07/ey-survey-c-suites-pivot-from-ai-adoption-to-unlocking-value-as-escalating-token-costs-trigger-fiscal-scrutiny", "canonicalUrl": "https://www.ey.com/en_us/newsroom/2026/07/ey-survey-c-suites-pivot-from-ai-adoption-to-unlocking-value-as-escalating-token-costs-trigger-fiscal-scrutiny", "sourceType": "INDEPENDENT_REPORTING", "sourceTier": 3, "publicationDate": "2026-07-28", "retrievedAt": "2026-09-28T10:00:11.298Z", "lastVerifiedAt": "2026-10-01T08:01:57.893Z", "verificationStatus": "UNVERIFIED", "isPrimarySource": false, "supportsObservedFact": false } ] }, { "id": "AE-2026-008", "title": "Deloitte legal survey finds expected automation, insourcing and pricing-model pressure", "slug": "deloitte-legal-ai-pricing-pressure-survey", "eventDate": "2026-07-09", "publicationDate": "2026-07-09", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Deloitte Legal" ], "sectors": [ "legal", "consulting" ], "sectorLabels": [ "legal", "consulting" ], "summary": "Deloitte’s survey of 121 senior legal leaders found 61% in AI deployment phases, 61% experimenting with or piloting agentic AI, 78% wanting external-provider AI to reduce costs, and 85% expecting AI to change law-firm pricing.", "observedFact": "Deloitte’s survey of 121 senior legal leaders found 61% in AI deployment phases, 61% experimenting with or piloting agentic AI, 78% wanting external-provider AI to reduce costs, and 85% expecting AI to change law-firm pricing. Respondents expected hourly work to fall materially over two to three years.", "interpretation": "Strong evidence that reduced human progression is being connected by buyers to pricing redesign, insourcing and lower dependence on external legal labour.", "mechanisms": [ "HBP", "CSD" ], "laws": [ "LAW_2", "LAW_3" ], "propositions": [ "REVENUE_DURABILITY" ], "themes": [ "outcome-pricing" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "MISALIGNED", "predictionTested": "When structured legal production requires less human progression, billable-hour durability should come under pressure while judgement and accountability remain more defensible.", "whyItMatters": "The evidence connects adoption to expected change in the commercial model as well as to productivity.", "counterInterpretation": "Survey expectations may not translate into realised fee changes.", "whatWouldStrengthen": "Observed declines in hourly share, realised alternative-fee growth and measurable insourcing.", "whatWouldWeaken": "If AI automates significant work but hourly billing remains dominant and economically durable for several years.", "relatedEvidenceIds": [ "AE-2026-018", "AE-2026-022", "AE-2026-041" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.509Z", "sources": [ { "id": "f672846d-255e-4042-a1a8-cb752c87cb97", "evidenceEventId": "AE-2026-008", "sourceName": "Deloitte UK", "headlineAtVerification": "AI set to reshape legal work, law firm pricing and legal careers | Deloitte UK", "sourceUrl": "https://www.deloitte.com/uk/en/about/press-room/ai-set-to-reshape-legal-work-law-firm-pricing-and-legal-careers.html", "canonicalUrl": "https://www.deloitte.com/uk/en/about/press-room/ai-set-to-reshape-legal-work-law-firm-pricing-and-legal-careers.html", "sourceType": "INSTITUTIONAL_RESEARCH", "sourceTier": 3, "publicationDate": "2026-07-09", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:51.941Z", "verificationStatus": "VERIFIED_INSTITUTIONAL", "isPrimarySource": false, "supportsObservedFact": true } ] }, { "id": "AE-2026-007", "title": "Bank of England records rapid gains in agentic task capability and governance constraints", "slug": "bank-of-england-agentic-capability-governance", "eventDate": "2026-07-07", "publicationDate": "2026-07-07", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Bank of England" ], "sectors": [ "financial-services" ], "sectorLabels": [ "financial-services" ], "summary": "The July Financial Stability Report said frontier models can sustain longer, more complex multi-step tasks with less human intervention, while also warning that agentic workflows can be costly and that autonomous payments raise questions around authorisation, traceability, liability and governance.", "observedFact": "The July Financial Stability Report said frontier models can sustain longer, more complex multi-step tasks with less human intervention, while also warning that agentic workflows can be costly and that autonomous payments raise questions around authorisation, traceability, liability and governance.", "interpretation": "Strong support for the direction of HBP reduction and Law 3, combined with a genuine boundary condition, since reliability, cost and governance can delay economically usable autonomy.", "mechanisms": [ "HBP" ], "laws": [ "LAW_3" ], "propositions": [], "themes": [ "governance", "verification", "boundary-conditions", "agentic-commerce" ], "classification": "BOUNDARY", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "This is an anticipated timing and reliability boundary already allowed by the paper. It is recorded because it affects the pace of commercially usable autonomy. Its existence does not weaken the framework.", "predictionTested": "Technical maturity changes the timing of commercial pressure, while governance and control become more important as autonomous execution expands.", "whyItMatters": "An independent central bank is observing both rising autonomy and the constraints that determine when it becomes economically deployable.", "counterInterpretation": "Financial stability analysis focuses on risk and may over-represent high-consequence use cases relative to the wider economy.", "whatWouldStrengthen": "Observed financial-sector deployments where autonomous progression grows and governance becomes a measurable bottleneck or cost centre.", "whatWouldWeaken": "If agentic costs and reliability limitations persist long enough to prevent meaningful deployment.", "relatedEvidenceIds": [ "AE-2026-005", "AE-2026-006", "AE-2026-015", "AE-2026-020" ], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.440Z", "sources": [ { "id": "89d244c6-f56f-486d-9edd-89f04498daf2", "evidenceEventId": "AE-2026-007", "sourceName": "Bank of England", "headlineAtVerification": "Financial Stability Report - July 2026 | Bank of England – the UK's central bank", "sourceUrl": "https://www.bankofengland.co.uk/financial-stability-report/2026/july-2026", "canonicalUrl": "https://www.bankofengland.co.uk/financial-stability-report/2026/july-2026", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-07-07", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:51.758Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-006", "title": "J.P. Morgan describes corporate treasury as an autonomous real-time control system", "slug": "jp-morgan-autonomous-treasury-control-system", "eventDate": "2026-06-25", "publicationDate": "2026-06-25", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "J.P. Morgan" ], "sectors": [ "financial-services" ], "sectorLabels": [ "financial-services" ], "summary": "J.P.", "observedFact": "J.P. Morgan described corporate treasury evolving into a real-time AI control system that can sense, predict, decide, execute and audit, while explicitly raising the question of accountability when agents act.", "interpretation": "Evidence that financial progression can move toward autonomous execution while judgement, accountability and control rise in relative importance.", "mechanisms": [ "HBP" ], "laws": [ "LAW_1", "LAW_3" ], "propositions": [], "themes": [ "governance", "judgement", "orchestration" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "Routine progression can become autonomous while scarce judgement and responsibility concentrate above the execution layer.", "whyItMatters": "The architecture described is close to the paper’s predicted separation between autonomous throughput and scarce supervisory judgement.", "counterInterpretation": "The piece is strategic thought leadership and does not show broad production deployment.", "whatWouldStrengthen": "Named production deployments with measurable reductions in manual treasury progression and explicit governance structures.", "whatWouldWeaken": "If autonomous treasury remains advisory-only or requires essentially unchanged human execution.", "relatedEvidenceIds": [ "AE-2026-005", "AE-2026-007", "AE-2026-015", "AE-2026-020" ], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.372Z", "sources": [ { "id": "5a909c67-2072-41fa-8e5d-c2d7e863392d", "evidenceEventId": "AE-2026-006", "sourceName": "J.P. Morgan", "headlineAtVerification": "Agentic AI in Corporate Cash & Treasury Management", "sourceUrl": "https://www.jpmorgan.com/payments/newsroom/agentic-ai-corporate-cash-treasury-management", "canonicalUrl": "https://www.jpmorgan.com/payments/newsroom/agentic-ai-corporate-cash-treasury-management", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-06-25", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:55.189Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-045", "title": "UK consumers retain purchase control when using AI shopping agents", "slug": "aci-uk-consumers-retain-purchase-control", "eventDate": "2026-06-22", "publicationDate": "2026-06-29", "discoveredDate": "2026-09-30", "datePrecision": "OBSERVED_BY", "companies": [ "ACI Worldwide" ], "sectors": [ "commerce" ], "sectorLabels": [ "commerce" ], "summary": "Post-publication UK consumer survey showing a strong distinction between AI assistance and delegated purchasing authority.", "observedFact": "YouGov surveyed 2,080 UK adults for ACI Worldwide from 19 to 22 June 2026. 19% trusted AI to make everyday purchasing decisions under user-set rules, 69% did not, 70% said purchases made without asking would affect willingness to use an AI shopping agent, and 60% said one mistake would cause them to stop using one.", "interpretation": "Boundary evidence for HBI. Agentic discovery can expand while consumers retain approval over economically consequential decisions.", "mechanisms": [ "HBI" ], "laws": [], "propositions": [], "themes": [ "agentic-commerce", "human-approval", "consumer-trust" ], "classification": "BOUNDARY", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "Retained human control at consequential transaction points is an anticipated trust and control boundary. The monitor will track whether this persists or recedes as trusted delegation infrastructure develops.", "predictionTested": "Tests whether autonomous systems displace human interaction at transaction and decision stages.", "whyItMatters": "The fieldwork occurred after publication and directly measures willingness to delegate purchase authority.", "counterInterpretation": "This is stated preference and does not record transaction behaviour. Better trust and payment controls may increase delegation over time.", "whatWouldStrengthen": "Observed transaction data showing human confirmation remains common as agentic commerce scales.", "whatWouldWeaken": "Large-scale autonomous purchasing without per-transaction approval.", "falsificationRelevance": "If scaled agentic commerce continues to require human approval at the transaction stage, HBI should be qualified by consequence and authority.", "relatedEvidenceIds": [], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-30T17:47:30.758Z", "createdAt": "2026-09-30T17:47:30.758Z", "updatedAt": "2026-10-11T09:24:49.805Z", "sources": [ { "id": "25a7f3b9-6790-42e8-8e5f-2f25bdac56b1", "evidenceEventId": "AE-2026-045", "sourceName": "ACI Worldwide", "headlineAtVerification": "Six in Ten UK Consumers Would Stop Using an AI Shopping Agent After One Mistake, ACI Survey Finds", "sourceUrl": "https://investor.aciworldwide.com/news-releases/news-release-details/six-ten-uk-consumers-would-stop-using-ai-shopping-agent-after", "canonicalUrl": "https://investor.aciworldwide.com/news-releases/news-release-details/six-ten-uk-consumers-would-stop-using-ai-shopping-agent-after", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-06-29", "retrievedAt": "2026-09-30T17:47:30.758Z", "lastVerifiedAt": "2026-09-30T17:47:30.758Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-005", "title": "KPMG and Microsoft deploy Agent 365 governance across global AI agents", "slug": "kpmg-microsoft-agent-365-governance", "eventDate": "2026-06-09", "publicationDate": "2026-06-09", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "KPMG", "Microsoft" ], "sectors": [ "consulting", "accounting", "software" ], "sectorLabels": [ "consulting", "accounting", "software" ], "summary": "KPMG and Microsoft announced global deployment of Agent 365 and Copilot, including tooling to manage, monitor and secure AI agents across KPMG and client organisations, alongside Copilot deployment to more than 276,000 professionals.", "observedFact": "KPMG and Microsoft announced global deployment of Agent 365 and Copilot, including tooling to manage, monitor and secure AI agents across KPMG and client organisations, alongside Copilot deployment to more than 276,000 professionals.", "interpretation": "Strong evidence that governance and control are becoming core production architecture required to scale autonomous throughput.", "mechanisms": [ "HBP" ], "laws": [ "LAW_1", "LAW_3" ], "propositions": [], "themes": [ "governance", "verification", "orchestration" ], "classification": "SUPPORTIVE", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "predictionTested": "As autonomous throughput expands, value and organisational importance move toward monitoring, control, verification and governed execution.", "whyItMatters": "Governance is being operationalised as infrastructure for enterprise-scale agent deployment.", "counterInterpretation": "The investment could reflect transitional enterprise risk management, and may not mark a durable economic layer.", "whatWouldStrengthen": "Evidence that governance platforms become budgeted, revenue-generating, mission-critical control layers tied to autonomous scale.", "whatWouldWeaken": "If governance becomes trivial, commoditised or economically negligible once agents mature.", "relatedEvidenceIds": [ "AE-2026-006", "AE-2026-007", "AE-2026-015", "AE-2026-020" ], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.288Z", "sources": [ { "id": "47336b62-899c-41bb-abff-e7c08ae20db5", "evidenceEventId": "AE-2026-005", "sourceName": "Microsoft", "headlineAtVerification": "KPMG and Microsoft scale trusted, enterprise AI agents globally through deployment of Agent 365 and Copilot - Source", "sourceUrl": "https://news.microsoft.com/source/2026/06/09/kpmg-and-microsoft-scale-trusted-enterprise-ai-agents-globally-through-deployment-of-agent-365-and-copilot/", "canonicalUrl": "https://news.microsoft.com/source/2026/06/09/kpmg-and-microsoft-scale-trusted-enterprise-ai-agents-globally-through-deployment-of-agent-365-and-copilot/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-06-09", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:55.850Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-004", "title": "Worldline, ING and Mastercard complete live end-to-end European agentic payment", "slug": "mastercard-worldline-ing-live-agentic-payment", "eventDate": "2026-06-02", "publicationDate": "2026-06-02", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Mastercard", "Worldline", "ING" ], "sectors": [ "financial-services", "commerce" ], "sectorLabels": [ "financial-services", "commerce" ], "summary": "The firms announced Europe’s first live end-to-end agentic payment transaction in production, with an AI agent initiating an authenticated payment using existing payment infrastructure.", "observedFact": "The firms announced Europe’s first live end-to-end agentic payment transaction in production, with an AI agent initiating an authenticated payment using existing payment infrastructure.", "interpretation": "Direct evidence that human-bounded interaction can be compressed as agents move from discovery or assistance into transaction initiation.", "mechanisms": [ "HBI" ], "laws": [ "LAW_1", "LAW_3" ], "propositions": [], "themes": [ "agentic-commerce", "interface-disintermediation", "governance" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "predictionTested": "Agents can mediate economic interaction and execute transactions without requiring the user to navigate each conventional interface step.", "whyItMatters": "The HBI mechanism is observable in production as well as in product concepts.", "counterInterpretation": "A single production transaction does not establish material adoption or revenue displacement.", "whatWouldStrengthen": "Meaningful transaction volumes, merchant adoption, recurring autonomous purchasing and observable changes to discovery or payment economics.", "whatWouldWeaken": "If agentic payments remain niche pilots requiring nearly the same human interaction as conventional transactions.", "relatedEvidenceIds": [ "AE-2026-043" ], "paradoxRelevance": "EXCLUDED", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.211Z", "sources": [ { "id": "99e07251-67ed-42d8-bf0e-f86fc9c07f67", "evidenceEventId": "AE-2026-004", "sourceName": "Mastercard", "headlineAtVerification": "Worldline, ING and Mastercard complete a live end-to-end European agentic payment in production | Mastercard Newsroom", "sourceUrl": "https://newsroom.mastercard.com/news/europe/en/newsroom/press-releases/en/2026/worldline-ing-and-mastercard-complete-a-live-end-to-end-european-agentic-payment-in-production/", "canonicalUrl": "https://newsroom.mastercard.com/news/europe/en/newsroom/press-releases/en/2026/worldline-ing-and-mastercard-complete-a-live-end-to-end-european-agentic-payment-in-production/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-06-02", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:55.633Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-043", "title": "Visa survey finds only 23% trust GenAI to make payments on their behalf", "slug": "visa-consumer-trust-agentic-payments-human-control", "eventDate": "2026-05-28", "publicationDate": "2026-09-09", "discoveredDate": "2026-09-30", "datePrecision": "OBSERVED_BY", "companies": [ "Visa" ], "sectors": [ "commerce" ], "sectorLabels": [ "commerce" ], "summary": "Visa's Trust Index found that only 23% of US consumers trusted generative AI to handle payment transactions on their behalf. Trust rose to 61% when Visa was named as the party securing the transaction. The survey was fielded from 26 to 28 May 2026, after publication of Autonomy Economics.", "observedFact": "In a Harris Poll survey fielded for Visa from 26 to 28 May 2026 among 2,065 US adults, only 23% trusted generative AI to handle payment transactions on their behalf. For the payment-transaction question, trust rose to 61% when Visa was the named payment brand securing the transaction.", "interpretation": "This is a boundary condition for Human-Bounded Interaction. Consumers may be willing to use AI for discovery and recommendation while retaining human or trusted-institution control at the point where money moves. It suggests that interface bypass can progress unevenly across the transaction journey and that delegated payment authority may remain more constrained than upstream shopping activity.", "mechanisms": [ "HBI" ], "laws": [], "propositions": [], "themes": [ "agentic-commerce", "consumer-trust", "human-approval", "payments" ], "classification": "BOUNDARY", "evidenceStrength": 3, "chronology": "POST_PUBLICATION", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "Retained human control at consequential transaction points is an anticipated trust and control boundary. The monitor will track whether this persists or recedes as trusted delegation infrastructure develops.", "predictionTested": "Tests whether autonomous systems displace human interaction at economically consequential transaction points, as well as during search and recommendation.", "whyItMatters": "The evidence is useful because it measures consumer willingness after the framework was published and focuses on the payment decision itself. It narrows any reading of HBI that assumes rapid removal of human approval throughout the commerce journey.", "counterInterpretation": "The survey measures stated trust and does not observe purchasing behaviour. Trust also rises substantially when a recognised payments institution is involved, which may indicate that stronger trust infrastructure can enable agentic transactions instead of preventing them.", "whatWouldStrengthen": "Observed transaction data showing persistent human confirmation requirements as agentic shopping volumes scale across multiple markets.", "whatWouldWeaken": "A sustained increase in autonomous transactions without human approval, especially where consumers delegate standing payment authority at scale.", "falsificationRelevance": "If agent-mediated commerce grows substantially while consumers continue to retain approval at the transaction stage, the scope and pace of HBI would need to be narrowed.", "relatedEvidenceIds": [ "AE-2026-004" ], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-30T17:06:51.127Z", "createdAt": "2026-09-30T17:06:51.127Z", "updatedAt": "2026-10-11T09:24:49.674Z", "sources": [ { "id": "db5fc1c3-a57b-44ab-aae3-56991cda8427", "evidenceEventId": "AE-2026-043", "sourceName": "Visa", "headlineAtVerification": "New Visa Research Finds Consumer Trust is Accelerating the Path to Agentic Commerce", "sourceUrl": "https://investor.visa.com/news/news-details/2026/New-Visa-Research-Finds-Consumer-Trust-is-Accelerating-the-Path-to-Agentic-Commerce/default.aspx", "canonicalUrl": "https://investor.visa.com/news/news-details/2026/New-Visa-Research-Finds-Consumer-Trust-is-Accelerating-the-Path-to-Agentic-Commerce/default.aspx", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-09-09", "retrievedAt": "2026-09-30T17:06:51.127Z", "lastVerifiedAt": "2026-09-30T17:06:51.127Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-003", "title": "Salesforce reports strong growth while agentic architecture expands", "slug": "salesforce-q1-fy27-agentic-growth", "eventDate": "2026-05-27", "publicationDate": "2026-05-27", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Salesforce" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Salesforce reported record first-quarter results, more than $1 billion in Agentforce ARR, $3.4 billion in combined AI and data ARR, 3.8 billion Agentic Work Units delivered, and more than one million active Slack MCP users within six weeks of launch.", "observedFact": "Salesforce reported record first-quarter results, more than $1 billion in Agentforce ARR, $3.4 billion in combined AI and data ARR, 3.8 billion Agentic Work Units delivered, and more than one million active Slack MCP users within six weeks of launch.", "interpretation": "Important boundary and comparative evidence. An incumbent can remain economically strong while agentic use rises, which is consistent with the paper if the firm is redesigning its seat-and-interface model instead of preserving it unchanged.", "mechanisms": [ "HBI" ], "laws": [ "LAW_1", "LAW_2" ], "propositions": [ "REVENUE_DURABILITY", "COMMERCIAL_ALIGNMENT" ], "themes": [], "classification": "BOUNDARY", "evidenceStrength": 4, "chronology": "POST_PUBLICATION", "commercialAlignment": "ALIGNED", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "Incumbent strength during redesign is consistent with the paper's distinction between deployment strength and business-model durability. The observation is tracked comparatively and does not count as weakening merely because the incumbent remains strong.", "predictionTested": "Firms that redesign around autonomous activity can preserve durability even while the old interaction structure weakens.", "whyItMatters": "It prevents the theory from being misread as 'AI must cause SaaS decline' and provides a live comparator for redesign versus non-redesign.", "counterInterpretation": "Salesforce may simply be benefiting from scale, installed-base strength and bundling, with agentic redesign playing little part.", "whatWouldStrengthen": "Sustained growth alongside increasing non-seat monetisation, headless use and autonomous activity.", "whatWouldWeaken": "If agentic products grow but remain economically immaterial to revenue durability or if redesign fails to preserve margins and retention.", "relatedEvidenceIds": [ "AE-2026-010", "AE-2026-011", "AE-2026-012", "AE-2026-013" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.104Z", "sources": [ { "id": "3f221ca3-59b2-4f45-ab78-f388c599f1af", "evidenceEventId": "AE-2026-003", "sourceName": "Salesforce", "headlineAtVerification": "Salesforce Delivers Record First Quarter Fiscal 2027 Results - Salesforce", "sourceUrl": "https://www.salesforce.com/news/press-releases/2026/05/27/fy27-q1-earnings/", "canonicalUrl": "https://www.salesforce.com/news/press-releases/2026/05/27/fy27-q1-earnings/?bc=DB", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-05-27", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:55.503Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-002", "title": "OpenAI launches Deployment Company with major capital partners", "slug": "openai-deployment-company", "eventDate": "2026-05-11", "publicationDate": "2026-05-11", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "OpenAI", "TPG", "Advent", "Bain Capital", "Brookfield", "Tomoro" ], "sectors": [ "private-equity", "software", "consulting" ], "sectorLabels": [ "private-equity", "software", "consulting" ], "summary": "OpenAI launched the OpenAI Deployment Company to embed forward-deployed engineers in organisations, agreed to acquire Tomoro, and described a model for scaling deployment across the economy with capital and consulting partners.", "observedFact": "OpenAI launched the OpenAI Deployment Company to embed forward-deployed engineers in organisations, agreed to acquire Tomoro, and described a model for scaling deployment across the economy with capital and consulting partners.", "interpretation": "A direct example of a central AI capability and deployment layer designed to generalise implementation knowledge across many organisations.", "mechanisms": [], "laws": [ "LAW_1" ], "propositions": [ "CAPITAL_STRUCTURED_AUTONOMY" ], "themes": [ "orchestration", "absorption" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "CONTEMPORANEOUS", "commercialAlignment": "ALIGNED", "predictionTested": "Value can move above individual operating companies toward actors that coordinate autonomous deployment, learning and workflow redesign across wider networks.", "whyItMatters": "The organisation is explicitly designed around repeatable deployment capability, and model access is only one part of it.", "counterInterpretation": "The company may function principally as a scaled professional-services arm for OpenAI.", "whatWouldStrengthen": "Evidence of repeatable cross-company deployment patterns, shared playbooks and measurable portfolio-level performance improvements.", "whatWouldWeaken": "If deployments remain bespoke and economic value continues to reside primarily in model access, with little moving to deployment and orchestration.", "relatedEvidenceIds": [ "AE-2026-001" ], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:48.020Z", "sources": [ { "id": "5234c5ae-428c-44a5-ac91-ca36ead4b59d", "evidenceEventId": "AE-2026-002", "sourceName": "OpenAI", "headlineAtVerification": "OpenAI launches the OpenAI Deployment Company to help businesses build around intelligence | OpenAI", "sourceUrl": "https://openai.com/index/openai-launches-the-deployment-company/", "canonicalUrl": "https://openai.com/index/openai-launches-the-deployment-company/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-05-11", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-09-04T23:07:58.720Z", "verificationStatus": "BROKEN_LINK", "isPrimarySource": true, "supportsObservedFact": true, "verificationNotes": "Did not resolve during the monthly source check." } ] }, { "id": "AE-2026-001", "title": "Anthropic and private-equity consortium form enterprise AI services company", "slug": "anthropic-private-equity-consortium-ai-services-company", "eventDate": "2026-05-04", "publicationDate": "2026-05-04", "discoveredDate": "2026-09-04", "datePrecision": "EXACT", "companies": [ "Anthropic", "Blackstone", "Hellman & Friedman", "Goldman Sachs", "General Atlantic", "Leonard Green", "Apollo Global Management", "GIC", "Sequoia Capital" ], "sectors": [ "private-equity", "software", "consulting" ], "sectorLabels": [ "private-equity", "software", "consulting" ], "summary": "Anthropic, Blackstone, Hellman & Friedman and Goldman Sachs announced a new AI services company for mid-sized enterprises.", "observedFact": "Anthropic, Blackstone, Hellman & Friedman and Goldman Sachs announced a new AI services company for mid-sized enterprises. Anthropic applied-AI engineers are to work alongside the company’s engineering team, while a wider consortium of asset managers backs the vehicle.", "interpretation": "Evidence of capital and ownership networks being used as a coordination layer for shared autonomous capability and deployment across multiple firms.", "mechanisms": [], "laws": [ "LAW_1" ], "propositions": [ "CAPITAL_STRUCTURED_AUTONOMY" ], "themes": [ "orchestration", "absorption" ], "classification": "SUPPORTIVE", "evidenceStrength": 5, "chronology": "CONTEMPORANEOUS", "commercialAlignment": "ALIGNED", "predictionTested": "Capital owners can organise deployment, engineering capability, standards and learning across networks of firms, instead of each operating company adopting independently.", "whyItMatters": "The structure closely matches the paper’s predicted ecosystem-level orchestration model.", "counterInterpretation": "It may prove to be a conventional AI services distribution company, which would leave the case for an enduring capital-coordination model unproven.", "whatWouldStrengthen": "Evidence that shared implementation patterns, governance, infrastructure or lessons are repeatedly reused across portfolio companies and produce measurable gains.", "whatWouldWeaken": "If deployments remain effectively one-off consulting engagements with little cross-company reuse or coordination.", "relatedEvidenceIds": [ "AE-2026-002" ], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "createdAt": "2026-09-04T00:00:00.000Z", "updatedAt": "2026-10-11T09:24:47.920Z", "sources": [ { "id": "27836342-6adf-4e18-a21b-57a1c424488f", "evidenceEventId": "AE-2026-001", "sourceName": "Anthropic", "headlineAtVerification": "Building a new enterprise AI services company with Blackstone, Hellman & Friedman, and Goldman Sachs \\ Anthropic", "sourceUrl": "https://www.anthropic.com/news/enterprise-ai-services-company?pubDate=20260225", "canonicalUrl": "https://www.anthropic.com/news/enterprise-ai-services-company?pubDate=20260225", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-05-04", "retrievedAt": "2026-09-04T00:00:00.000Z", "lastVerifiedAt": "2026-10-01T08:01:58.084Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-040", "title": "Microsoft prices Agent 365 at $15 per user despite the product governing autonomous agents", "slug": "microsoft-agent-365-retains-per-user-pricing", "eventDate": "2026-05-01", "publicationDate": "2026-05-01", "discoveredDate": "2026-09-30", "datePrecision": "EXACT", "companies": [ "Microsoft" ], "sectors": [ "software" ], "sectorLabels": [ "software" ], "summary": "Microsoft made Agent 365 generally available as a control plane for AI agents while retaining a per-user commercial unit. The standalone product is priced at $15 per user per month, with each licence covering an individual who manages or sponsors agents, or uses agents to do work on their behalf.", "observedFact": "On 1 May 2026 Microsoft made Agent 365 generally available as a standalone licence at $15 per user per month. Microsoft states that each licence covers an individual who manages or sponsors agents, or uses agents to do work on their behalf.", "interpretation": "This is a boundary condition for Law 2. A product built specifically to govern autonomous agents can still use a human access unit when the accountable sponsor or user remains a meaningful commercial denominator. It shows that autonomous activity does not make seat or user pricing disappear immediately, especially in control-plane products.", "mechanisms": [], "laws": [ "LAW_2" ], "propositions": [], "themes": [ "access-pricing", "agent-governance", "control-plane", "pricing-boundary" ], "classification": "BOUNDARY", "evidenceStrength": 3, "chronology": "CONTEMPORANEOUS", "commercialAlignment": "UNCLEAR", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "The paper explicitly allows access-based models to persist for a period. This record is useful because it shows one form that persistence can take while autonomous activity expands; it is not by itself evidence that Law 2 has weakened.", "predictionTested": "Tests whether access-based pricing becomes commercially marginal as autonomous activity expands.", "whyItMatters": "The observation is useful because the product sits close to the autonomous control layer and still remains priced per user. It narrows any overly simple reading of Law 2 that would expect agentic products to abandon user-based pricing immediately.", "counterInterpretation": "This is a vendor pricing decision and does not show that customers prefer or will retain the model over time. The licence is attached to the accountable human sponsor instead of to each agent, so it may be consistent with value concentrating in governance while still using a human billing denominator.", "whatWouldStrengthen": "Evidence that Agent 365 or comparable control-plane products sustain per-user pricing and pricing power over several years as agent volumes scale.", "whatWouldWeaken": "A move by Microsoft or comparable vendors towards action, usage or outcome pricing as autonomous volumes rise, or evidence that customers resist per-user licensing for agent control planes.", "falsificationRelevance": "If per-user pricing remains durable across agentic control-plane products as autonomous activity scales, it would weigh against a strong near-term reading of Law 2.", "relatedEvidenceIds": [], "paradoxRelevance": "CONTEXTUAL", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-30T16:07:16.128Z", "createdAt": "2026-09-30T16:07:16.128Z", "updatedAt": "2026-10-11T09:24:49.468Z", "sources": [ { "id": "e5997b2b-1528-4817-9da0-0665e591cde2", "evidenceEventId": "AE-2026-040", "sourceName": "Microsoft", "headlineAtVerification": "Microsoft Agent 365, now generally available, expands capabilities and integrations", "sourceUrl": "https://www.microsoft.com/en-us/security/blog/2026/05/01/microsoft-agent-365-now-generally-available-expands-capabilities-and-integrations/", "canonicalUrl": "https://www.microsoft.com/en-us/security/blog/2026/05/01/microsoft-agent-365-now-generally-available-expands-capabilities-and-integrations/", "sourceType": "PRIMARY", "sourceTier": 1, "publicationDate": "2026-05-01", "retrievedAt": "2026-09-30T16:07:16.128Z", "lastVerifiedAt": "2026-10-01T08:01:58.744Z", "verificationStatus": "VERIFIED_PRIMARY", "isPrimarySource": true, "supportsObservedFact": true } ] }, { "id": "AE-2026-041", "title": "Most law firms report unchanged pricing despite clients expecting AI-driven commercial change", "slug": "thomson-reuters-law-firm-pricing-unchanged-ai", "eventDate": "2026-04-30", "publicationDate": "2026-06-22", "discoveredDate": "2026-09-30", "datePrecision": "OBSERVED_BY", "companies": [], "sectors": [ "legal" ], "sectorLabels": [ "legal" ], "summary": "Thomson Reuters' 2026 Future of Professionals survey, conducted in March and April before the Autonomy Economics publication date, found a clear gap between client expectations and law-firm pricing response. Seventy-one per cent of in-house legal professionals expected outside firms to change their commercial models as AI usage increased, while only 28% of law firms reported pricing changes and 62% said pricing remained unchanged. Because the underlying survey predates publication, this record is treated as contemporaneous background and does not count as post-publication evidence.", "observedFact": "In the 2026 Future of Professionals legal findings, 71% of in-house legal professionals say they expect outside firms to change their commercial models as AI usage increases. Thomson Reuters reports that only 28% of law firms have made changes to pricing in response to AI, while 62% of law-firm professionals say their firm's pricing structure is unchanged.", "interpretation": "This is useful contemporaneous boundary evidence for Human-Bounded Progression and Law 2, and does not count as a post-publication test. It shows that commercial structures can lag client expectations and AI adoption. The record remains in the register because it helps define the boundary, but it must not contribute to claims about what happened after the framework was published.", "mechanisms": [ "HBP" ], "laws": [ "LAW_2" ], "propositions": [ "COMMERCIAL_ALIGNMENT" ], "themes": [ "legal-pricing", "billable-hour", "client-pressure", "commercial-lag" ], "classification": "BOUNDARY", "evidenceStrength": 4, "chronology": "CONTEMPORANEOUS", "commercialAlignment": "MISALIGNED", "signalStage": "EARLY_SIGNAL", "boundaryType": "ANTICIPATED", "horizonNote": "This observation shows that commercial structures can lag changing expectations. That persistence is useful to track now, while the one-to-two-year commercial-unit durability test remains open.", "predictionTested": "Tests whether AI-enabled changes in professional work translate into changes in effort-based commercial units and whether commercial structures adapt alongside activity structures.", "whyItMatters": "It provides direct survey evidence that client expectations and law-firm pricing structures are currently out of alignment. That gap is exactly where a boundary observation is useful. Pressure is observable while redesign has not become dominant.", "counterInterpretation": "The survey measures expectations and reported pricing changes. It does not measure realised reductions in human progression or long-run revenue durability. Law firms may be adapting through rates, staffing, scope or service mix without changing the formal pricing structure.", "whatWouldStrengthen": "Firm-level data linking realised AI productivity to hours, staffing, pricing structure and client retention over several years.", "whatWouldWeaken": "Evidence that law firms sustain unchanged hourly structures, margins and client retention for several years after substantial realised reductions in human progression.", "falsificationRelevance": "If unchanged effort-based pricing remains economically durable for several years after measurable reductions in human progression, this would move from boundary evidence towards the framework's stated falsification condition.", "relatedEvidenceIds": [ "AE-2026-008", "AE-2026-018", "AE-2026-022" ], "paradoxRelevance": "CORE", "publicStatus": "PUBLISHED", "reviewedBy": "editor", "reviewedAt": "2026-09-30T16:07:16.128Z", "createdAt": "2026-09-30T16:07:16.128Z", "updatedAt": "2026-10-11T09:24:49.537Z", "sources": [ { "id": "f82085be-40de-4dc8-9ebc-452e15bed7ed", "evidenceEventId": "AE-2026-041", "sourceName": "Thomson Reuters Institute", "headlineAtVerification": "Future of Professionals - 2026 Legal Report", "sourceUrl": "https://www.thomsonreuters.com/en/institute/future-of-professionals-2026/report-legal", "canonicalUrl": "https://www.thomsonreuters.com/en/institute/future-of-professionals-2026/report-legal", "sourceType": "INSTITUTIONAL_RESEARCH", "sourceTier": 3, "publicationDate": "2026-06-22", "retrievedAt": "2026-09-30T16:07:16.128Z", "lastVerifiedAt": "2026-10-01T08:01:59.108Z", "verificationStatus": "VERIFIED_INSTITUTIONAL", "isPrimarySource": false, "supportsObservedFact": true } ] } ], "chains": [ { "id": "governance-and-verification", "title": "Governance and verification", "description": "Control, authorisation and accountability moving from compliance afterthought to production architecture and systemic-risk concern.", "eventIds": [ "AE-2026-005", "AE-2026-006", "AE-2026-007", "AE-2026-015", "AE-2026-020" ] }, { "id": "software-durability-capital-markets", "title": "Software Revenue Durability and capital markets", "description": "Whether perceived durability under autonomy transmits into financing conditions before revenue resets.", "eventIds": [ "AE-2026-019", "AE-2026-021", "AE-2026-035" ] }, { "id": "legal-commercial-unit-pressure", "title": "Legal commercial-unit pressure", "description": "From expected pricing change, to buyers actively pressing the billable hour, to the first evidence that billed hours are rising instead of falling while AI use is near-universal.", "eventIds": [ "AE-2026-008", "AE-2026-018", "AE-2026-022", "AE-2026-041" ] }, { "id": "salesforce-commercial-redesign", "title": "Salesforce commercial redesign", "description": "Whether an incumbent can preserve Revenue Durability by redesigning its commercial architecture as autonomous activity grows.", "eventIds": [ "AE-2026-003", "AE-2026-010", "AE-2026-011", "AE-2026-012", "AE-2026-013" ] }, { "id": "capital-structured-autonomy", "title": "Capital-Structured Autonomy", "description": "Capital ownership organising autonomous capability, deployment and learning across networks of firms.", "eventIds": [ "AE-2026-001", "AE-2026-002" ] }, { "id": "agentic-commerce-hbi", "title": "Agentic commerce and interface displacement", "description": "Agents moving from assistance into transaction initiation. Awaiting production-volume evidence.", "eventIds": [ "AE-2026-004", "AE-2026-043" ] } ], "statusHistory": [ { "propositionKey": "HBP", "status": "STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-005", "AE-2026-006", "AE-2026-007", "AE-2026-008", "AE-2026-014", "AE-2026-018" ] }, { "propositionKey": "HBI", "status": "STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-003", "AE-2026-004", "AE-2026-010", "AE-2026-011", "AE-2026-012", "AE-2026-016", "AE-2026-019", "AE-2026-021" ] }, { "propositionKey": "CSD", "status": "STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-008", "AE-2026-009", "AE-2026-014", "AE-2026-016", "AE-2026-019", "AE-2026-020", "AE-2026-021" ] }, { "propositionKey": "LAW_1", "status": "VERY_STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-001", "AE-2026-002", "AE-2026-003", "AE-2026-004", "AE-2026-005", "AE-2026-006", "AE-2026-009", "AE-2026-010", "AE-2026-011", "AE-2026-012", "AE-2026-016", "AE-2026-020" ] }, { "propositionKey": "LAW_2", "status": "VERY_STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-003", "AE-2026-008", "AE-2026-012", "AE-2026-013", "AE-2026-014", "AE-2026-016", "AE-2026-017", "AE-2026-018" ] }, { "propositionKey": "LAW_3", "status": "VERY_STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-004", "AE-2026-005", "AE-2026-006", "AE-2026-007", "AE-2026-008", "AE-2026-011", "AE-2026-013", "AE-2026-014", "AE-2026-015", "AE-2026-017", "AE-2026-018", "AE-2026-020" ] }, { "propositionKey": "AGENTIC_PROFIT_PARADOX", "status": "MODERATE", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-014", "AE-2026-018" ] }, { "propositionKey": "CAPITAL_STRUCTURED_AUTONOMY", "status": "DIRECTLY_OBSERVABLE", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-001", "AE-2026-002" ] }, { "propositionKey": "REVENUE_DURABILITY", "status": "STRONG", "direction": "UNCHANGED", "assessedAt": "2026-09-04", "reason": "Initial seed register baseline, 1 May to 4 September 2026", "eventIds": [ "AE-2026-003", "AE-2026-008", "AE-2026-012", "AE-2026-014", "AE-2026-018", "AE-2026-019", "AE-2026-021" ] } ], "reviews": [ { "reviewedAt": "2026-09-06", "kind": "MANUAL", "sourcesChecked": 22, "sourcesFailed": 0, "statusChanges": 0, "note": "Register reviewed during deployment; all sources resolved." }, { "reviewedAt": "2026-10-01", "kind": "MONTHLY", "sourcesChecked": 31, "sourcesFailed": 8, "statusChanges": 0, "note": "8 source(s) did not resolve and need review." }, { "reviewedAt": "2026-10-01", "kind": "MONTHLY", "sourcesChecked": 31, "sourcesFailed": 8, "statusChanges": 0, "note": "8 source(s) did not resolve and need review." } ], "falsification": [ { "quarter": "2026-Q4", "reviewedAt": "2026-10-01", "propositionKey": "CSD", "strongestContraryEvidence": "The Business Insider piece on AI's \"productivity paradox\" (June 2026) reports that despite massive corporate AI spending and demonstrable speed gains at task level, with engineers completing week-long tasks in days, \"AI productivity gains haven't transferred to areas beyond coding as neatly\" and \"researchers say AI gains have yet to consistently translate into greater company productivity, revenue, or profits.\" Critically, this data covers the claimed one-to-two-year test horizon and encompasses the period after autonomous capability has been \"credibly deployed in commercially meaningful ways\" (coding, cybersecurity). The article also notes companies are still ramping up AI integration and facing an \"automation phase\" requiring \"upfront investment\" before \"the bigger unlock\" materialises. This suggests that commercial structures dependent on labour scarcity, in domains where AI capability has genuinely expanded access, are not experiencing margin pressure or forced redesign so far, even as capability has demonstrably broadened. The source explicitly states the expected pathway (productivity boost → cost savings → margin defense) has not materialized at company level within this window, despite task-level capability expansion.", "sourceUrls": [ "https://www.businessinsider.com/companies-waiting-ai-productivity-boom-2026-6", "https://www.businessinsider.com/ai-jobs-crisis-no-evidence-apollo-chief-economist-torsten-slok-2026-5", "https://www.businessinsider.com/openai-won-ai-compute-race-business-model-next-anthropic-2026-4" ], "genuinelyContradicts": false, "alternativeReading": "Commercial structures dependent on capability scarcity do face exposure as autonomous capability expands, but the translation from task-level capability expansion to commercial durability erosion is lengthened by integration costs, deployment overhead, and uneven maturity across domains. The structures are adapting instead of staying durable and unchanged. They absorb productivity gains into reinvestment cycles in place of margin capture, which postpones the visible squeeze on scarcity-dependent pricing. Once integration matures and offsetting costs normalize, pressure on HBP/HBI/CSD models should follow. The one-to-two-year horizon may be too compressed for sectors with deep integration requirements or multi-layered proprietary dependencies.", "recommendation": "QUALIFY", "reasoning": "The evidence presents a current counter-signal and falls short of a completed falsification. The framework claims that once autonomous capability is credibly deployed in commercially meaningful ways, \"revenue models calibrated to HBP, HBI or CSD should come under pressure\" within one to two years. HBP is human-bounded progression, HBI is human-bounded interaction and CSD is capability scarcity dependence. The June 2026 article directly addresses this horizon. Capability (AI-assisted coding, data analysis) has been deployed commercially and is measurably expanding access. The expected pressure on scarcity-dependent structures, meaning margin erosion and forced business model redesign, has still not materialised. Instead, companies report increased spending, extended timelines, and delayed payoff. \n\nHowever, this does not falsify the framework, for four reasons. (1) The paper anticipated \"uneven capability maturity\" and integration depth as boundaries. Coding shows clear wins; other domains show \"woefully inadequate\" performance due to training data scarcity (per the Palo Alto Networks CEO in the September article). (2) The same article documents companies in an active \"automation phase\" buildout, the upfront integration cost phase the framework did not prohibit. (3) The June article acknowledges workers are producing \"more code than ever,\" indicating access-broadening is real. Durability appears to come from implementation investment costs absorbing the productivity boost. (4) No claim of frontier premiums has been tested so far, because capability maturity varies sharply across domains.\n\nThe counter-signal is that scarcity-dependent labour structures in high-capability domains (coding, cybersecurity) are not under margin pressure after about 18 months of commercial deployment, contrary to the predicted one-to-two-year trigger. For this to become framework-level disconfirmation, capability would need to remain deployed and mature (and not degraded), access would need to remain broad or cost-substitutable, and commercial structures would need to hold margins and revenue models unchanged for another 6 to 12 months without redesign.\n\nThe framework's anticipated boundaries (uneven maturity, integration depth, cost of deployment) are present in the data. The counter-signal does not falsify; it highlights that the causal pathway is slower or less direct than the mechanism implies at this horizon." }, { "quarter": "2026-Q4", "reviewedAt": "2026-10-01", "propositionKey": "HBI", "strongestContraryEvidence": "The strongest evidence against the claim comes from the customer service sector, where autonomous agents are operationally deployed but interaction-dependent revenue models persist with pricing power intact. The Business Insider article on AI agents calling customer-service lines (Sept 2026) documents that (1) AI callers \"remain a small share of total volume,\" (2) call centre agents continue to be employed and graded on performance metrics, (3) the existing revenue model, where companies monetise customer support staff, shows no sign of redesign or margin pressure, and (4) functional boundaries persist, as in \"only the account holder can close their credit card\" and \"only the customer can do, so they have to be the one on the line.\" After credible commercial deployment in a sector where interaction mediation should theoretically reduce HBI dependence most directly, the core revenue structure (labour-intensive customer service scaled by interaction volume) remains structurally intact without pressure to redesign. This is the operational counter-signal at the horizon the test specifies.", "sourceUrls": [ "https://www.businessinsider.com/ai-agents-customer-service-calls-companies-2026-9" ], "genuinelyContradicts": true, "alternativeReading": "The mechanism may operate selectively across sectors instead of as a general principle. Autonomous systems can mediate interaction in domains where regulatory and authentication boundaries are weak (real estate search, security operations coordination, internal workforce tasks). But in sectors where transaction authority, legal liability, or regulatory compliance requires retained human touch-points, the displacement of interaction-dependent revenue may be structurally limited. The claim would then hold for high-interaction-mediation sectors (e.g., research navigation, task scheduling) but not for transaction authority sectors (financial services, insurance claims, medical authorization). This would make the claim narrower than stated, contingent on where decision authority rests compared with information-gathering authority.", "recommendation": "QUALIFY", "reasoning": "The claim predicts that where \"value capture depends on sustained human interaction, autonomous systems can reduce or displace that interaction.\" The test specifies that \"revenue models calibrated to HBP, HBI or CSD should come under pressure\" once autonomous capability is \"credibly deployed in commercially meaningful ways.\" \n\nThe customer service evidence documents exactly this condition. Autonomous agents are deployed (Meta's Muse, Instinct, specialised law bots), deployed commercially (multiple sectors, including insurance, credit cards and attorney firms), and deployed repeatedly at scale (one agent calling \"a couple times every shift\"). The incumbent revenue model, which monetises labour-intensive customer interaction, still persists without redesign. Call centres remain staffed, employees are graded on metrics, and compensation and utilisation remain tied to call handling. No credible evidence appears of margin pressure, price reduction, headcount cuts, or model redesign at the institutions receiving these autonomous calls.\n\nHowever, the evidence also documents built-in functional boundaries that the paper anticipated as potential limits, including \"only the account holder can close their credit card,\" authentication requirements, and retained human approval. These are anticipated boundaries in the paper's framework and are not new revelations. \n\nThe counter-signal is therefore genuine but scoped. In transaction authority domains, regulation and law keep sustained human interaction structurally required, whatever the state of the technology or its cost. This suggests the mechanism operates differentially by sector. The claim needs qualification. It holds for information-discovery and navigation tasks (where agents can act end-to-end), but interaction-dependent revenue models in high-liability, authority-critical transactions remain durable for reasons beyond those the claim identifies. If this boundary persists through the test horizon (into 2027 and 2028), it moves from a current counter-signal to a framework-level limitation on the claim's universality." }, { "quarter": "2026-Q4", "reviewedAt": "2026-10-01", "propositionKey": "HBP", "strongestContraryEvidence": "The strongest case against the claim comes from the Census Bureau's microstructure study (Bonney et al., April 2026). Among firms actively deploying AI across multiple business functions and making operational investments, employment does decline, but this effect disappears once you control for functional deployment breadth and operational investment. More importantly, 66% of firms use AI purely for task augmentation with no employment reductions (2% report cuts), and worker-task use shows no association with employment decline after controlling for deployment scope. This suggests human progression remains embedded in firms' economic models even where autonomous capability is deployed commercially. The Agent-Centric Enterprise paper (January 2026) reports 92% reduction in audit reporting time but does not claim that audit departments have been eliminated or that revenue models calibrated to human effort have been redesigned, only that the work takes less time. The Demirer et al. \"Chaining Tasks\" paper (February 2026) confirms that AI steps cluster into chains but emphasises that firms still bundle steps into human jobs and must navigate coordination costs. It predicts reorganisation and does not predict a loosening of human progression dependence. The Babina et al. paper (August 2026) finds productivity gains driven by organisation capital accumulation from AI-skilled jobs, meaning human learning and firm-specific knowledge, which is precisely human-bounded progression. None of these sources report sectors in which autonomous capability has been credibly deployed commercially while revenue models remain calibrated to human progression and durability holds with expanding margins.", "sourceUrls": [ "https://www.nber.org/system/files/working_papers/w35141/w35141.pdf", "https://hdsr.mitpress.mit.edu/pub/0mrfxamu", "https://www.nber.org/system/files/working_papers/w34859/w34859.pdf", "https://www.nber.org/system/files/working_papers/w35684/w35684.pdf" ], "genuinelyContradicts": false, "alternativeReading": "The evidence supports a modified reading. Autonomous capability can reduce the time, effort, or cognitive load required per unit of work within established business processes (task augmentation), while firms have not redesigned revenue models away from human-bounded progression so far. Instead, they are layering autonomous tools onto human-centered workflows and capturing short-term efficiency gains without structural change. The durability of revenue models calibrated to human progression is not being tested, because firms have not faced pressure to justify value through that lens once capability is deployed. The mechanism may operate, but over a longer horizon than the current 12 to 24 months of deployment, or only after a trigger event forces explicit model redesign.", "recommendation": "QUALIFY", "reasoning": "The claim itself, that work can be carried forward with less dependence on continuous human progression, is directionally supported by the evidence. Task times are falling (92% audit report reduction), firms are deploying autonomous capability commercially (18 to 22% of US firms, and 60 to 70% of large information and finance firms), and the Census data confirms this is real. However, the paper's specific test mechanism, that revenue models calibrated to human-bounded progression should come under pressure within one to two years, finds no clear disconfirmation and no confirmation either. The evidence suggests a different sequence. Firms are achieving productivity gains through augmentation and organisational capital without redesigning the commercial models that justified the human progression in the first place. The census data shows deployment breadth is still narrow (57% of adopters use AI in 3 or fewer functions; 65% restrict worker use to 3 or fewer tasks), employment effects are muted, and augmentation dominates. This is consistent with the \"embedding distribution and switching costs\" boundary conditions the paper already anticipated. The current counter-signal is that durability is holding precisely because firms have not been forced to rebuild workflows as agent-centric systems (as Demirer et al. and the Agent-Centric Enterprise paper argue must happen). Revenue models calibrated to human progression remain intact because the autonomous capability has been bolted on and has not transformed the model. If this persists for one to two more years despite deployment maturity, it weakens the claim's mechanism. Right now it qualifies the claim, since the mechanism may be real but operating on a longer or conditional timeline than stated." } ], "audits": [ { "slug": "2026-09", "title": "Autonomy Economics Evidence Audit for September 2026", "periodStart": "2026-09-01", "periodEnd": "2026-09-30", "strengthened": [ "Salesforce explicitly prices Help Agent on successful outcomes (strength 5)", "Wall Street banks push Big Law to pass AI productivity into lower fees (strength 5)", "FT reports consulting clients cutting external work and pressing fees as AI expands internal capability (strength 5)", "Salesforce and Anthropic announce Claudeforce (strength 5)", "Salesforce expands Headless 360 so agents can invoke enterprise capabilities directly (strength 5)", "The Economist reports that China's AI drive threatens the world's largest workforce (strength 5)", "EY survey finds C-suites pivoting from AI adoption to unlocking value as token costs trigger fiscal scrutiny (strength 5)", "Worldline, ING and Mastercard complete live end-to-end European agentic payment (strength 5)", "OpenAI launches Deployment Company with major capital partners (strength 5)", "Anthropic and private-equity consortium form enterprise AI services company (strength 5)", "Blackstone links realised AI productivity gains to early repricing and a shift from seats to outcomes (strength 4)", "Atlassian prices customer-service AI at $1 per successful resolution (strength 4)", "FCA finds enterprise value depends increasingly on the AI harness around the model (strength 4)", "Atlassian introduces metered Rovo credits for AI and Teamwork Graph usage (strength 4)", "FSB elevates autonomous-AI control and resilience into G20 financial-system discussion (strength 4)", "DeepSeek open-sources modular agent harness (strength 4)", "Deloitte legal survey finds expected automation, insourcing and pricing-model pressure (strength 4)", "J.P. Morgan describes corporate treasury as an autonomous real-time control system (strength 4)", "KPMG and Microsoft deploy Agent 365 governance across global AI agents (strength 4)" ], "weakened": [], "counterSignals": [], "boundaryConditions": [ "Salesforce Q2 results show strong durability during commercial and architectural redesign (anticipated boundary)", "Bank of England records rapid gains in agentic task capability and governance constraints (anticipated boundary)", "UK consumers retain purchase control when using AI shopping agents (anticipated boundary)", "Salesforce reports strong growth while agentic architecture expands (anticipated boundary)", "Most law firms report unchanged pricing despite clients expecting AI-driven commercial change (anticipated boundary)", "Futurum survey finds buyers still prefer per-seat pricing for separately billed generative AI (anticipated boundary)", "Visa survey finds only 23% trust GenAI to make payments on their behalf (anticipated boundary)", "Microsoft prices Agent 365 at $15 per user despite the product governing autonomous agents (anticipated boundary)" ], "tooEarly": [], "eventIds": [ "AE-2026-013", "AE-2026-018", "AE-2026-014", "AE-2026-011", "AE-2026-010", "AE-2026-024", "AE-2026-023", "AE-2026-004", "AE-2026-002", "AE-2026-001", "AE-2026-035", "AE-2026-017", "AE-2026-022", "AE-2026-020", "AE-2026-019", "AE-2026-016", "AE-2026-015", "AE-2026-044", "AE-2026-012", "AE-2026-009", "AE-2026-008", "AE-2026-007", "AE-2026-006", "AE-2026-045", "AE-2026-005", "AE-2026-003", "AE-2026-041", "AE-2026-021", "AE-2026-042", "AE-2026-043", "AE-2026-040" ], "publishedAt": "2026-09-06T21:57:51.583Z" } ] }