From Autonomy Economics, the reference entry for the discipline
Autonomy changes the economics of activity, interaction, and capability access. In many settings, activity has long depended on HBP across stages of interpretation, decision, and execution, interaction on sustained HBI in how demand is formed and captured, and capability access on CSD where value depends on scarce underlying capability. Revenue structures are the focus here because profit pressure builds through their weakening. When the commercial basis on which revenue is justified erodes, the profit structure that depends on it comes under pressure as well. This does not unfold in a linear sequence. It can appear wherever systems start to carry forward parts of that activity, mediate parts of that interaction, or make underlying capability more accessible in commercially meaningful ways.
The examples here draw on software, services, financial institutions, wealth management, consulting, private equity, and private credit. They illustrate a broader pattern without claiming to define its full boundaries.
Human contribution remains central, but its role changes as autonomous systems carry forward parts of activity and interaction, and as access to capability broadens. The balance between contribution and progression adjusts, and the place of interaction and capability access within the commercial model changes with it. Pressure often appears early, before adoption is complete, because expectations can change as soon as credible alternatives come into view. Firms also face rising costs of deploying, integrating, and governing these systems.
As activity is carried forward differently, interaction takes place differently, and access to capability is reorganised, revenue models tied to earlier forms of progression, interaction, or scarcity can come under pressure. This reflects a change in alignment between the structure of activity and the structure of value capture.
Software spans several revenue structures, and exposure varies within the category. Some products earn revenue from seats, others from usage, tokens, compute, or API calls. Some earn revenue from cycles such as runs, scans, or jobs. Many combine software with services. Exposure depends on the structure of the revenue line and the role the product plays in how activity is carried forward, how interaction takes place, or how access to capability is organised.
Seat-based models may weaken when activity can be carried forward without occupying a seat in the same way. Products that package fixed activity behind a user interface may be among the most exposed. As systems start to carry that activity forward more directly, the marginal value of an additional seat can decline. The continued role of human developers does not, on its own, preserve the earlier commercial structure. Developers using agentic coding tools still produce and deliver software, but the activity may no longer depend on HBP in the same way. The economic question moves from time spent to systems delivered, integration quality, and outcome performance.
Usage-based models can strengthen or weaken. Autonomous activity can increase usage when the product remains central to how activity is carried forward or interaction is mediated. It can reduce usage when systems bypass part of that path. Durability depends on whether the product remains embedded in the activity structure and whether it continues to sit inside the interaction path through which demand is routed or executed.
Cycle-based models earn revenue from runs, scans, builds, or jobs. Autonomy can generate or direct more of these cycles. If the product remains the environment in which they run, usage can rise. When those cycles can be executed elsewhere, usage can fall.
Hybrid models face pressure from more than one direction. The software side reflects how autonomy interacts with seats, usage, or cycles. The services side reflects how autonomy interacts with progression. In some cases, pressure can also appear in the capability layer when the technical or analytical capability on which the offer depended becomes easier to access elsewhere. Pressure can therefore appear across several layers at once, although some firms may strengthen their position when they redesign around autonomous capability.
Some products gain value by embedding autonomy directly. They move from packaging activity to enabling capability and may strengthen when they become part of the environment in which autonomous execution occurs. Pressure can still appear when firms built on top of third-party autonomous capability lose the scarcity that supported their position. This can happen when the underlying capability becomes more widely accessible, more easily substituted, or is absorbed by the upstream provider within its own product environment. In such cases, downstream firms may retain users and continue improving their products, and still face pressure on the commercial basis of value capture because the capability layer on which they depended is no longer scarce on the same terms.
Other products may weaken when autonomous systems can generate or execute the underlying activity without them. In these cases, progression moves outside the product and the revenue line becomes less durable. Where the product also depended on repeated interface use or navigation, interaction-based pressure may appear alongside progression pressure.
Some products become infrastructure for autonomy. They provide data, coordination, integration, or execution environments that autonomous systems depend on. Their value is tied less to bounded interaction and more to enabling progression and interaction at scale. In some cases, they also benefit because they become one of the environments through which previously scarce capability is operationalised.
Some software businesses derive part of their value from access to technical capability that remains difficult, costly, or complex for others to reproduce. As autonomous capability improves and becomes easier to access through models, agents, tooling layers, or open systems, that scarcity can weaken. In these cases, profit pressure has a further source beyond activity moving outside the product or interaction being displaced. The underlying capability on which the product’s value depended becomes less scarce and less defensible.
Across software, the pattern is broadly consistent. Revenue tends to hold when the product remains central to how activity is carried forward, how interaction is mediated, or how access to capability is structured. It tends to weaken when those functions can move outside it.
Some services businesses earn revenue from activity that progresses through human involvement, including analysis, interpretation, planning, drafting, and coordination. These activities have long relied on human judgement to carry progression forward. As autonomy carries more of this activity, the basis on which value is justified can become less stable. The link between progression and human effort weakens.
Other services also depend on interaction. Value may be tied to the work itself and also to discovery, access, client engagement, and the relationship surface through which work is formed and sold. Where autonomous systems reduce the need for continuous interaction of that kind, pressure may appear even if the underlying service still matters.
In other cases, the service depends in part on scarce analytical, technical, or operational capability that clients previously could not easily access on their own. As that capability broadens, scarcity weakens and the commercial basis of the service may come under pressure even before demand falls.
Other services operate on a different footing. Their value rests more firmly on judgement, accountability, trust, relationship depth, or physical presence. In these settings, autonomy supports delivery while authority and responsibility continue to anchor the commercial logic.
This distinction separates services exposed through progression, interaction, or scarcity from those grounded more firmly in authority and responsibility.
Banks and asset managers may show the same pattern in operational, compliance, research-heavy, and interface-dependent layers. Fee pools tied to progression can begin to soften when parts of that progression move with less continuous dependence on HBP. Research, onboarding, monitoring, and reconciliation have long depended on HBP to interpret information and determine what happens next. As autonomy advances these activities, both the cost base and the logic supporting those fee pools can come under pressure.
Interaction also matters. In some parts of financial services, value depends on controlling the client interface, information flow, or discovery process through which products are selected or actions are initiated. Where autonomous systems begin to mediate that interaction more directly, pressure may build on the surface through which value was previously captured.
Capability scarcity matters as well. Some parts of the model depend on access to technical, cognitive, or operational capability that once remained difficult to reproduce outside specialised teams. As that capability becomes more accessible in commercially meaningful ways, parts of the revenue base may become less secure.
In regulated settings, verification, supervision, and accountability carry more weight as volume and speed increase. The more stable parts of the model tend to sit closer to judgement, control, responsibility, and governed use.
Wealth management may show a slower version of the same pattern. Adviser judgement, trust, behavioural guidance, and responsibility remain central in complex client settings.
Pressure can appear in adjacent activities such as portfolio monitoring, research preparation, reporting, and elements of portfolio construction. As autonomous capability advances in these areas, some clients may question the relationship between effort and value. Demand for advice can remain strong while parts of the process no longer appear as scarce, and some of the analytical capability on which they depended becomes easier to access.
Interaction matters here too. Wealth management has often depended on sustained client contact, relationship maintenance, and the advisory surface through which confidence is built and decisions are framed. Where some of that interaction becomes easier to mediate through autonomous systems, pressure may appear on parts of the model that were supported by that relationship layer rather than by final judgement alone.
Regulatory pressure reinforces this across multiple jurisdictions. Firms are increasingly required to demonstrate that the value clients receive is justified by outcomes they understand and benefit from. The activity involved in producing those outcomes is no longer accepted as the justification. The result is pressure from more than one direction. Autonomous capability affects the commercial basis for certain revenue structures, while regulation raises the standard for how that value is demonstrated.
Consulting firms face direct exposure. A large share of billable activity is tied to analysis, synthesis, and structured problem solving. Autonomous systems can carry these activities forward quickly. They do not need to replace entire engagements to affect the model. A change in how clients assess the cost of producing the underlying activity can be enough.
Interaction matters here as well. Some consulting value has long depended on workshops, discovery processes, iterative client engagement, and the interface through which problems are shaped and reframed. Where parts of that interaction can be compressed, routed differently, or handled with less continuous human involvement, pressure may appear there too.
Scarcity also matters. In some cases, the pressure sits in the weakening scarcity of analytical and synthetic capability that clients can increasingly access more directly. Judgement, political navigation, implementation support, and senior trust remain valuable, but they may carry more of the defensible premium as underlying analytical capability broadens.
Firms that redesign early can use lower-cost analysis and preparation to widen coverage, improve responsiveness, and deepen senior attention. The advantage depends on how additional capacity is converted into stronger commercial position.
Legal and accounting activity contains several progression-heavy layers. Drafting, review, checking, and reconciliation have long depended on HBP across defined stages. As autonomous capability becomes more useful in these areas, pressure often appears first in the parts of the model tied to progression, while final responsibility is less exposed.
Exposure is uneven. Structured progression may come under pressure earlier. Judgement, accountability, sign-off, and governed interpretation retain more economic weight. In legal and accounting work, capability scarcity can also matter. Some of the technical and analytical capability that once sat behind specialised teams may become easier to access for clients and adjacent providers, even where final responsibility remains with licensed or accountable professionals.
Interaction also remains relevant in some parts of the model, particularly where discovery, client contact, and the management of process have historically formed part of value capture. That pressure is likely to be less immediate than progression pressure in many cases, but it may still appear in adjacent layers.
An early signal of this pressure appears in audit and advisory pricing, where clients are already seeking a share of AI-enabled efficiency gains and regulators are simultaneously pressing firms to strengthen governance and verification around AI use.
Private equity and private credit face the issue from more than one direction.
Within portfolio companies, autonomous capability can improve operations and strengthen earnings. At the same time, many of those companies generate revenue from software or services tied to progression, interaction, or capability access. As their customers adopt autonomy, willingness to pay for those structures can weaken.
Private credit shows a related pattern. Monitoring, servicing, underwriting support, and reporting can increasingly be carried forward with less human involvement. This can improve internal efficiency while also affecting how parts of the model are commercially justified. In some parts of the market, interface control and information access may also matter, especially where the lender or platform sits in the path through which decisions, monitoring, or distribution take place.
The effect can appear across operating performance, revenue durability, and valuation at the same time.
Across these categories, a consistent pattern appears. Demand can remain intact while pressure builds. Capability improves and operations strengthen, while the commercial structure can begin to weaken.
In each of the cases examined, revenue lines built on HBP, HBI, or CSD may become less secure once those conditions can be carried forward, mediated, or accessed in different ways.
The changes above are calibrated to the current 2026 picture rather than assuming full substitution. Adoption is broadening, but it remains uneven, and the areas seeing the clearest pressure are still the ones where capability is reliable enough to matter commercially, governance can support deployment, and clients can already see a different basis for pricing or value.
From section 8, Where Profits Get Eaten, of the foundation paper. HBP, HBI and CSD stand for human-bounded progression, human-bounded interaction and capability scarcity dependence, the three axes of exposure.