Small businesses may turn out to be where enterprise AI becomes real first.

Not because they have bigger budgets, better infrastructure or dedicated AI teams, but because they sit much closer to the work itself. They know what a missed lead costs, what an extra pair of hands is worth, and exactly where time disappears. As AI shifts from software that helps people work to software that does the work, that proximity changes the equation.

The question is no longer whether SMEs will adopt AI. It’s whether AI gives them access to capabilities that once required a much larger organisation, and whether that changes what it means to be small in the first place.

Sources: BlackRock — The Machine-Native Economy · McKinsey — Is That AI Agent Worth It? · Futurism — Workers Spending Their Own Money on AI · TechCrunch — Meta Launches Enterprise AI Platform · Engadget — Meta Starts an Enterprise AI Business · Seams — Don’t Price Per Token · Seams — AI Product Portal · FactoryJet — US Small Business AI Adoption · British Chambers of Commerce — Half of SMEs Using AI · NatWest — AI Growth Gap Across UK Businesses · YouGov — Are UK Business Decision Makers Embracing AI? · YouGov — UK SME Leaders on AI Adoption

Google is building infrastructure designed to run billions of autonomous agent workloads. At the same time, researchers are warning that human oversight becomes harder as autonomy increases, while new model architectures like Jev’s are adding specialised decision layers inside agents themselves. Google AX · AI Agents Push Humans Out of the Loop · LangChain on Jev

What we call “an agent” is starting to look less like a single intelligence and more like an architecture of intelligences; reasoning, routing, classifying, governing and acting at different speeds.

That creates a new AX problem: how do you make all of that observable without making it overwhelming?

The answer may not be keeping humans in every loop, but designing systems that know which loops need us, which don’t, and why.

Listen to today’s AX Report.

Buying intelligence is becoming easier. Turning it into dependable, profitable work is the business.

We used to buy software to help us do our work. We now buy work performed by software. Behind that change is another: companies can now acquire important parts of the intelligence required to deliver their products from external providers, then build a business around how that intelligence is applied.

Chatbase offers customer service agents using models from providers including OpenAI, Anthropic and Google. Its product brings those models together with company information, integrations and tools for deploying and managing agents. Retell combines language models, voice technology and telephony with its own infrastructure for running conversations. Both turn externally available capabilities into services customers can configure, deploy and pay to use. Chatbase, Retell

Software companies have long rented infrastructure and incorporated other companies’ technology. What changes here is the capability being acquired: intelligence that can interpret a request, make intermediate decisions and contribute to execution. Product teams will still have substantial engineering and operational work to do. But it can concentrate that investment on a particular customer problem rather than developing every underlying capability itself.

This changes where value is created and captured. Access to a capable model is available to competitors. Understanding the customer’s circumstances, coordinating the necessary systems and delivering useful work reliably require a more specific product and operating model.

The commercial opportunity lies in converting available intelligence into work that customers are willing to entrust to the product.

That is the territory of Agentic Experience. AX encompasses the agents, their operating conditions and the relationships through which people delegate work to them. Decisions about that arrangement determine what can be sold, what it costs to deliver and how much responsibility the product can credibly assume.

From selling access to supplying work
In a familiar SaaS model, a customer pays for employees to access software that helps them perform their jobs. A support platform, for example, gives a team the tools to manage enquiries, find information and record resolutions. Employees supply the judgement, effort and coordination required to resolve the customer’s problem. The subscription buys access to the application; the customer separately funds the people and processes that turn that access into completed work.

Agentic products bring more of that execution into the purchase. A customer can buy capacity to handle enquiries, apply policies and carry out authorised actions. People still establish objectives, supply context and manage exceptions, but the software assumes part of the workload. The billing format may remain a monthly subscription, with seats, usage allowances or platform fees. What changes is the responsibility the product takes on in return for that payment.

This changes the purchasing calculation. Customers must assess how much work the product completes, how quickly and to what standard, alongside the human effort still required. The subscription price becomes one component of the total cost of delivery, including setup, supervision, escalation and correction. A cheaper product that requires extensive checking may offer less value than a more expensive one that reliably completes the task.

It also changes how the provider can grow. A seat-based product typically earns more as additional employees gain access. An agentic product can earn more as customers delegate more work, even when the number of people supervising it stays the same. Agent design helps determine whether that growth is economical. An agent that recognises errors, recovers effectively and seeks human help at the appropriate point can reduce wasted execution and unnecessary supervision. These are core AX decisions: they shape how acquired intelligence becomes dependable work and what it costs to deliver.

The AX value chain
That transformation involves several stages. Model and infrastructure providers supply the underlying intelligence. The product combines it with the customer’s objective and relevant context, then equips agents with the tools and permissions needed to act. During execution, feedback allows agents to assess progress, correct mistakes or request human intervention. The result must then be checked against the task’s requirements. Value accumulates through this chain as general capability becomes completed work the customer can use.

Manus makes this progression tangible. Its Browser Operator can work within a customer’s existing browser session, accessing authorised research subscriptions and business systems to cross reference information, conduct analysis and compile reports. The customer authorises the session and can monitor, interrupt or take over execution. The product brings intelligence into the environment where the information and tools needed for the task already reside. Manus

Consider using that capability to prepare a competitor analysis. The brief establishes which companies and questions matter; access to subscribed research supplies context; browser actions gather the evidence; and synthesis turns it into a report. Verification then requires checking that the sources support the conclusions and that the report answers the brief. Producing a document alone does not establish success. Customer value comes from obtaining a useful analysis with less total effort, including the time spent directing and checking it. The commercial opportunity lies in making that whole sequence dependable enough to delegate.

Businesses can capture value at different points in this chain. OpenRouter, for example, provides access to external models through a common interface, with routing, usage tracking and fallbacks. It passes through provider inference prices and charges a fee when customers purchase credits; its published pay-as-you-go platform fee is 5.5%. Its commercial contribution lies in making intelligence easier to acquire and operate across suppliers. OpenRouter, OpenRouter

OpenRouter and Manus illustrate two different places to build a business in this value chain. OpenRouter charges for making external intelligence accessible and manageable. Manus packages intelligence with tools and an execution environment so customers can delegate tasks such as research and report preparation. For each, the commercial test is whether customers will pay more for the work delivered than it costs to acquire the intelligence, execute the task and handle exceptions.

AX shapes both sides of that equation through decisions about how work is requested, executed and checked. In the competitor analysis example, clarifying the brief before research begins can reduce irrelevant searches, unnecessary model calls and customer rework. Presenting conclusions alongside supporting sources can make the report easier to verify and reduce follow-up requests. Those decisions affect both how much effort the customer saves and how much the product spends delivering an acceptable result.

What’s the customer paying for?
The differences become tangible in the charging unit. A minute, a message credit and a completed resolution each represent a different commercial promise.

Retell charges for voice-agent usage by the minute. Its pricing separates components including voice infrastructure, text-to-speech and the language model. The customer is purchasing the capacity to run conversations, with costs affected by the chosen configuration and duration. Retell captures value through the infrastructure and services that make those conversations possible. Retell

Chatbase packages agent capacity into subscriptions. At its published monthly prices, the Standard plan includes 4,000 message credits for $150, with additional credits available through automatic recharge. Credit consumption varies by model, so an allowance does not necessarily correspond to the same number of responses in every configuration. The package combines access, functionality and a metered amount of execution. Chatbase, Chatbase

Decagon has described two alternatives for its customer service agents: charging per conversation, or charging a higher rate for a fully resolved conversation with no charge for escalations. In its published explanation, it reported that most customers preferred per-conversation pricing because it was easier to predict and avoided disputes about what qualified as a resolution. Decagon

Choosing between these mechanisms requires understanding the delegation itself. Can the customer reasonably predict how many minutes or messages a task will consume? Can they control that consumption without supervising every step? Can both parties recognise when the work is complete? AX makes these questions part of the pricing decision because they determine whether the charge feels understandable, controllable and justified.

Usage pricing is easier to justify when customers can select the configuration, anticipate consumption and set spending limits. It becomes harder to justify when an agent independently chooses how many steps to take and the customer pays for its retries. A credit allowance can make expenditure predictable, but customers still need to understand how much useful work it buys and what happens when it runs out. Outcome pricing can align payment with completed work, provided success can be clearly defined and verified. Otherwise, the product risks turning an unresolved customer need into a disputed bill.

The experience must therefore support the chosen commercial promise. Charging by usage requires visibility and budget controls. Charging by resolution requires evidence of completion and clear treatment of escalation, failure and reopened cases. These are product capabilities that make a pricing mechanism workable.

AX contributes to choosing the mechanism by establishing what responsibility the product can credibly assume and what uncertainty the customer is prepared to retain. The resulting price must cover delivery costs, but the charging unit must also reflect a delegation the customer understands and accepts.

AX decisions are margin decisions
Once execution consumes paid intelligence and external services, product behaviour affects delivery cost directly. An unnecessary question can extend a voice call. Poor context can trigger repeated attempts. A task routed to an unsuitable model can generate apparently inexpensive output that requires expensive correction.

The customer experiences these failures as delay, repetition or extra supervision. The provider experiences them as consumption, support demand or lost trust. Improving the experience can therefore improve the economics of delivery, provided the improvement reduces total effort rather than simply moving it elsewhere.

Consider a hypothetical service charging £2 for a completed task. If direct execution costs £0.40 and the average cost of retries and human exceptions adds another £0.60, £1 remains before other business expenses. If poor task definition causes those additional costs to double, that contribution falls to £0.40. The selling price has not changed. The ability to understand, execute and recover has changed the economics.

This makes the cost per accepted result more informative than the cost of an individual model response.

It includes the attempts that fail and the intervention required to make the work usable. On the customer’s side, the equivalent measure includes briefing, supervision, checking and correction. A product can be profitable for its provider while leaving the customer with too much effort to justify renewal.

AX connects customer value with delivery performance through practical decisions about how agents work. Devin’s interactive planning, for example, identifies relevant code, raises implementation questions and presents a plan that the user can adjust before coding begins. This creates an opportunity to correct misunderstandings before they become unnecessary execution and rework. Its documented test-generation workflow also returns proposed changes with a coverage summary, giving reviewers evidence to assess rather than simply declaring the task complete. Devin, Devin

Retell addresses the point where execution passes back to a person. Its warm-transfer feature can check that a human has answered and privately provide the receiving employee with caller context. That can spare customers from repeating themselves and reduce the employee’s effort in reconstructing the problem. These features show how clarification, verification and handover can improve the experience while reducing avoidable work. Their economic value depends on whether the effort they save exceeds the time and resources they add. Retell

Context and authority define delegation
A product’s commercial scope depends on what it knows and what it is authorised to do. Retrieving a policy, applying it to a customer’s circumstances and executing a decision represent progressively greater responsibility. AX makes that progression workable by establishing how context is maintained, authority is granted and exceptions return to human judgement.

GC AI’s Playbooks illustrate this in legal work, applying company-specific standards and preferred positions to contract review. The AX consideration is how legal teams establish those positions, keep them current and understand when a case requires their attention. Making those responsibilities practical can reduce repeated instruction and checking, increasing the amount of useful work the team can delegate. GC AI

Permissions determine how far that delegation can go. Too many approval requests return routine work to the customer; too few can expose them to decisions they never intended to delegate. AX connects agent behaviour with meaningful human control, allowing the product to assume greater responsibility where its context and reliability justify it. The commercial benefit is more work delivered with proportionate oversight and manageable risk.

Measuring AX
Experience work has often had to defend its budget through improvements in usability, satisfaction and engagement. When treated as a discretionary layer of quality, it becomes vulnerable to cuts.

With agentic products, decisions about the experience reach directly into the production and sale of work. How a task is understood, what an agent can execute and when a person must intervene affect the capacity the business can sell and the cost of supplying it.

AX’s contribution can therefore be measured across several dimensions. For customers, the measures include time to an acceptable result, quality of completion and effort remaining after delegation. For products, they include delivery cost, retries, human intervention and margin.

Commercially, the question is whether better delivery leads customers to renew, delegate more work or entrust the product with a broader range of responsibilities. Reliability and control matter throughout, growth that brings more unauthorised actions, disputed results or costly recovery may destroy the value it appears to create.

These measures need to be read together. Fewer escalations are valuable only if issues are still resolved correctly. Faster execution matters only if it does not increase checking and correction. More usage may represent stronger demand, or an agent struggling through repeated attempts. A defined task provides a practical basis for comparison, product teams can assess the same kind of work before and after a change, including its quality, total cost and human effort. That makes it possible to test whether an AX decision improved the economics rather than merely moving the burden.

The opportunity extends beyond efficiency. Clearer permissions can make a previously unacceptable task safe enough to delegate. Better context can enable a more specialised service. Stronger evidence of completion can support a commercial promise the provider could not previously make. AX generates business value when it improves existing delivery or expands the work the product can credibly undertake. Its contribution should be measured through those results.

Owning the promise
Those results give substance to the product promise. A customer buys an expectation that particular work will be completed to an agreed standard, within understood limits, with a clear response when something goes wrong. The product’s positioning, price, agent behaviour and human support must all sustain that expectation.

External intelligence makes it possible to assemble substantial capability without owning every component. It also leaves the business exposed to suppliers changing their models, prices and terms. The organisation must preserve the quality of its service through those changes. That requires ownership of context, evaluation, permissions and exceptions, alongside the ability to adjust how work is delivered. Renting capability does not transfer accountability for the customer’s result.

AX connects that operating responsibility to what the customer experiences. If an agent requires help, the handover must preserve the work already done. If performance falls short, the business must be able to recognise the failure and put it right. If customers entrust it with more consequential tasks, its controls and delivery capacity must support the greater responsibility. The experience is where the commercial promise is fulfilled, qualified or broken.

Practising AX therefore means following decisions through to their consequences for customers and the business. It requires connecting an understanding of human needs with agent behaviour, delivery economics and organisational responsibility and using evidence from completed work to improve all four.

As industrial strength intelligence becomes more widely available, the ability to organise it into a service worth buying becomes increasingly important.

The business value of AX lies in making more work worth delegating, more delivery economically sustainable and the product’s promise dependable enough to build a business around.

AX is the design of delegated work, how users express intent and retain control, how agents operate effectively, and how organisations deliver outcomes reliably and profitably. It builds on UX while connecting user experience, agent design and business accountability. As software assumes more responsibility, designers must help shape what it is authorised to do, who benefits and who bears the risk. AX expands the scope of design to match the scope of what is changing.

AX is not just UX for AI
Software is increasingly capable of undertaking work that people previously performed themselves. This changes what users delegate, what products take responsibility for and how businesses deliver value. For designers, it raises a reasonable challenge: why call this Agentic Experience? Doesn’t UX already address these questions? And where does designing the agents themselves fit?

User Experience (UX) concerns a person’s experience of a product or service: whether it meets their needs, how they understand and interact with it, and the effort, confidence and satisfaction involved. In agentic systems, this includes expressing intent, supervising work and assessing results. UX remains essential, with a scope that already extends beyond interfaces into the services and systems that shape people’s experiences.

Agentic Experience (AX), as proposed here, focuses on the whole system of delegated work. It encompasses the design of agents, the conditions in which they operate and the relationships through which people entrust work to them. It connects user needs with agent capability, delegated authority and organisational accountability, including how a business delivers that work reliably and profitably.

Within this scope, agent experience concerns the interfaces, instructions, APIs, tools, runtime environments, context and feedback that enable an agent to work effectively. These conditions influence whether it understands a task, takes an appropriate action and recognises success or failure. The quality of the agent’s experience directly affects the quality, cost and reliability of the work delivered to the user.

The AI native start-up Factory.AI makes these connections concrete. Its engineering agents have adjustable autonomy, approval requirements and organisational restrictions. Tools and execution feedback support the agent’s work; users need to understand its actions and manage the permissions available to them; organisational controls determine what authority can be delegated. The AX consideration is how these elements work together so that useful work is completed, oversight remains practical and accountability is clear. Factory

These are connected perspectives, not exclusive disciplinary territories. UX provides foundations for understanding and serving people. Agent design establishes capabilities and behaviour. Commercial and operational decisions determine what the product promises, how execution is supported and who bears the cost of failure. AX brings their interdependence into focus. A product may be easy to instruct yet expensive to supervise; an agent may execute effectively yet act beyond the authority its user intended to grant.

AX is therefore a proposed focus for an emerging practice. Its value must be demonstrated through better decisions about the work entrusted to software. For designers, that means participating in choices about what a product should undertake, what users should retain control over and how the resulting benefits and risks are distributed.

AX expands the scope of design to match the scope of what is changing. Making that expansion real requires influence over the products being built, the decisions shaping them and the organisations responsible for their delivery. Design has widened its remit before as technology changed the relationship between people and systems. That history provides a useful foundation for understanding what this next expansion demands.

We have been here before
The past 30 years of digital product design can be understood as a series of expansions in what designers consider their responsibility. These traditions overlap, and several began well before the web.

Human–computer interaction (HCI) established foundations for understanding how people work with computers, drawing on psychology, computing and human factors. User-centred design made people’s needs, abilities and circumstances central to the development process. Together, they helped establish that people should not have to accommodate every limitation imposed by a machine. University of London

As digital products became part of everyday life, user experience (UX) brought a broader relationship into view. The concern extended from completing an interaction to understanding the experience surrounding a product: expectations, usefulness, comprehension and satisfaction. UX was never intended to mean screens alone. Its established definition encompasses a person’s interactions with a company, its services and its products. The breadth of that ambition matters because AX builds on it. Nielsen Norman Group

Design thinking helped carry design approaches into organisational problem-solving, giving multidisciplinary teams ways to investigate needs, challenge assumptions and test possibilities. Frameworks such as the Double Diamond, shared by the Design Council from 2004, made that approach easier to communicate and adopt. Service design widened attention to the people, processes and operational arrangements behind an experience. A frustrating interaction could be understood as the consequence of a policy, a handover or an organisational boundary. Design Council

As digital delivery grew, design systems and design operations addressed another problem: how to sustain quality and consistency across many products and teams. Each development gave designers additional ways to influence what organisations produced. The underlying movement was towards a more complete understanding of the conditions that create an experience. These various methods became useful because they helped designers work at that expanding scale.

The work of design, the design of work
Agentic software brings another set of conditions into focus. When a system can undertake work, designers must help determine how responsibility passes between the person and the product.

Arcads, an AI-native platform for producing, editing, localising and scaling video adverts illustrates this in creative production. Its advertising agent, Mark, selects AI actors, assembles scenes, stitches clips and adds subtitles to produce finished adverts. Activities that previously required a sequence of production tasks can increasingly be delegated through a brief. Arcads

The creative team must still determine whether the result communicates the right message, represents the product accurately and is appropriate for its audience. Through an AX lens, the design challenge includes helping people express creative intent, compare alternatives and decide what is ready to publish.

Producing more material can reduce execution effort while increasing the burden of selection and review. The quality of delegation depends on how well the product supports the judgement that remains human.

A human-centred practice therefore remains the starting point. We need to understand what people want to delegate, what they need to control and what would give them confidence in the result. We must also consider people affected by the system who never directly configure it.

An AI customer-service agent acts for a business, but its behaviour affects the caller seeking help and the employee who may inherit an unresolved case. The person instructing the agent is only one of the people whose needs matter.

Retell AI, which provides infrastructure for the of production voice and chat agents, makes this wider relationship tangible. Its voice agents support warm transfers that confirm a human has answered and can privately provide the receiving employee with caller context before connecting the call. That capability addresses an important part of the experience: what happens when responsibility passes from software to a person. The design task includes determining when a transfer is appropriate and what the employee needs to know to continue effectively. Retell

This is where trust and empathy become operational. A reassuring voice has limited value if a caller cannot reach help or must repeat everything when they do. Trust depends on the system acting within understood boundaries and supporting recovery when it reaches them. Empathy requires attention to the whole arrangement: the person asking for help, the employee taking over and the information and authority each needs.

The business value of AX
The economics of that arrangement shape its human consequences. Retell’s voice-agent pricing is based on minutes, making usage cost and successful resolution distinct measures. For a business deploying such a system, the relevant economic question extends beyond the price of a call to the cost of resolving the customer’s need, including any escalation or repeat contact. This creates a design consideration: reducing conversation time is useful only if the customer still receives an adequate resolution. Retell

Creative production presents a related question. A team using Arcads may be able to generate more advertising variations with less production effort, but someone must still assess their suitability and decide which deserve testing. The economic benefit depends on the whole workflow, including briefing, generation, review, revision and approval. Output volume alone cannot tell us whether the team has gained productive capacity or simply moved its bottleneck.

Designers therefore need to understand how value is created, what delivery costs and where those costs fall. A product can save its provider money while increasing the customer’s effort. It can complete a task quickly while transferring the burden of verification to someone else. If routine work is delegated to agents, employees may inherit a concentrated workload of difficult exceptions. Pricing, performance measures and staffing decisions all influence the experience we are trying to improve.

Commercial understanding also makes design’s contribution more effective. Clearer intent can reduce unnecessary execution. Better review mechanisms can prevent costly errors. Appropriate boundaries can make a product dependable enough for customers to delegate more valuable work. Human needs and business viability meet in these decisions. It is the designer’s task is to make that relationship explicit and shape it deliberately.

Big design demands a big practice
Magic Patterns, a AI design and prototyping tool for product teams that turns text into visual interfaces and production-ready code, offers a useful example for designers wondering where familiar practices fit. It generates prototypes and can incorporate an existing design system. The design system consequently becomes a source of constraints for delegated production, alongside its role in helping people design and build consistently. But conformity to a component library does not establish that a generated design solves the right problem. Research, interaction design and critical judgement remain necessary to assess what the system produces. Magic Patterns

The same principle applies to other established design methods. A service blueprint can expose where an agent acts, where responsibility changes hands and where exceptions accumulate. Research can examine whether people understand the permissions they grant and can meaningfully supervise the resulting work. A design system can establish consistent ways to communicate uncertainty, request approval and show the status of an action. Existing practices can develop to address the new division of AX based labour.

Designers also need to participate where the underlying choices are made. That includes deciding what work a product should undertake, what constitutes an acceptable result, how much human review is necessary and what happens when execution fails. Factory’s controls make delegated authority explicit; Retell’s transfers support a return to human assistance; Arcads changes the relationship between production and creative judgement; Magic Patterns brings design systems into generated work. Each example connects an experience decision to the way work is organised.

We need enough fluency in models, evaluation, operating costs and organisational incentives to work effectively with engineering, operations, finance and risk. We earn that wider remit by helping resolve these questions. AX gives the work a focus: designing systems of delegation that people can use confidently and organisations can deliver responsibly.

Calling a few good designers
We need a vanguard of designers willing to take this broader responsibility into practice. Designers who can connect human needs to technical capability and commercial reality; who can question whether work should be delegated as carefully as they design how delegation happens; and who can help organisations distinguish productive automation from the transfer of effort and risk to someone else.

This is an opportunity to influence the design of work while its boundaries are being redrawn. The choices being made now will determine who exercises judgement, who retains control and how the benefits of increased capability are distributed. Design has a contribution to make wherever those choices affect people.

If AX expands the scope of design to match the scope of what is changing. Our responsibility is to make that expansion real, in the products we build, the decisions we influence and the organisations we help re/design.

.

TL;DR
The AX Shift moves software from tools people operate to systems they delegate work to. It changes what customers need and what they pay for, how providers deliver work profitably and how organisations divide responsibility between people and software. Product leaders must now re/design their products and businesses around reliable execution, human oversight and the economics of outcomes. AX is how they do it.

The AX Shift
The economics of software are changing. For the past 30 years, we bought software to help us do our work. Now we buy work performed by software.

You express an objective, establish boundary conditions and review the result. As more execution passes to the system, the basis of value changes. You start to judge a product by the work it completes, the supervision it requires and the outcome it delivers.

This transition from operating software to delegating work is the AX Shift.

AX, or Agentic Experience, is how agency is allocated between people and intelligent systems. It begins with understanding people’s needs, goals and circumstances. What do they want to delegate, what do they want to control, and what would give them confidence in the result? Designing for AX means helping people express intent, clarify misunderstandings and intervene when necessary. The economic value of delegation depends on that human experience. Software creates little benefit if supervising it is harder than doing the work.

These human needs shape the responsibilities a product can usefully assume, and the business required to deliver them. Transferring work to software also changes costs, accountability and customer expectations, extending AX beyond the traditional remit of interface, service and user experience design. Product leaders must now decide what their software will undertake, how people will direct and supervise it, how to deliver that work profitably and how to organise the business behind the promise.

The Big Boom
Since 2023 we’ve been experiencing the Big Boom, which is the rapid expansion of investment and valuations in AI, the proliferation of AI-powered software, and the growing range and complexity of work agentic systems can perform.

By the second half of 2026, the Big Boom was visible in the speed at which AI-native companies were building substantial businesses. Lovable, the AI application creation platform that turns natural-language instructions into functioning software, reported reaching a $500 million annualised revenue run rate in June. Sierra, which provides customer-service agents that resolve customer issues across channels, confirmed in August that it had reached $200 million ARR in nine quarters, doubling its earlier milestone in just two quarters. These companies sell different capabilities, software creation and customer service, but their growth demonstrates demand for products that undertake meaningful parts of the customer’s work. TechCrunch, Sierra

The scale extends beyond revenue. By August, Lovable reported more than 60 million projects created and over 900 million monthly visits to applications built on its platform. Its customer examples also reveal a potentially disruptive economic consequence, they claim one of their customers, Nursa, was retiring ten SaaS systems as teams built their own tools. While one company’s experience cannot establish a market-wide trend, it illustrates how delegated software creation can change the decision to build or buy, and put existing software revenues under pressure. Lovable

Investors are assigning extraordinary value to the intelligence suppliers behind this market. Anthropic’s May funding round valued it at $965 billion; OpenAI’s March round valued it at $852 billion. Those valuations encompass much more than coding, but coding agents provide concrete evidence of commercial demand: Claude Code had already exceeded $2.5 billion in annualised revenue by February. The economic significance lies in the work these products can absorb. As customers delegate implementation, service and other tasks, software companies can compete for spending previously committed to performing that work. The AX Shift is the challenge of turning that opportunity into dependable products and sustainable businesses. Anthropic, OpenAI, Claude Code

The capability to perform work is improving alongside that expansion. Stanford’s 2026 AI Index records agent success on WebArena’s realistic web tasks rising from approximately 15% in 2023 to 74.3% in early 2026.

These are benchmark results, not guarantees of production reliability, just as a generated project is not necessarily a viable product. But the strategic direction is significant, software is becoming easier to create while gaining the capacity to undertake more demanding tasks. As implementation becomes more accessible, competitive advantage increasingly depends on choosing valuable work, understanding its context and completing it dependably. Stanford AI Index 2026

Lovable illustrates both sides of this change. Its customers can describe an application and delegate much of its implementation, inspecting and steering the result as it develops. The product reduces the work required to turn an intention into functioning software. Its value therefore rests partly on the quality of the application produced and partly on how little effort the customer must expend getting there. Ease of expressing intent, resolving ambiguity and correcting direction becomes economically consequential: every failed interpretation consumes time and resources on both sides.

The product as work
When software undertakes a task, the product’s promise expands.

Cognition’s Devin, for example, allows developers to delegate engineering work and review the resulting pull request. Cognition’s guidance emphasises that engineers using Devin should adhere to detailed requirements, and follow their guidelines for testing and reviewing of the output. The economic question is consequently broader than whether the system can generate code. It is whether the engineering work it returns is useful enough to justify the combined cost of execution, supervision and correction. A fast result that requires extensive rework may create less value than a narrower task completed reliably. Cognition

This makes the customer’s remaining workload a central product measure. Delegation delivers value when the effort of specifying, supervising and checking the work is sufficiently smaller than the effort of doing it. Product teams must therefore design the whole relationship. How people communicate objectives, which decisions the system makes, when it asks for help and how results are accepted.

AX gives these decisions a common focus and forces teams to ask what responsibility the product can assume while making the customer’s work materially easier.

Different products will draw that boundary differently. Legora connects legal research, drafting and review while keeping lawyers in control of the resulting work. Its proposition accommodates a division in which execution can be delegated while professional judgement remains human. Retaining review does not diminish the product’s value if it concentrates a lawyer’s attention on decisions that require expertise. The appropriate level of agency is the one that produces useful work with an acceptable burden of oversight. Legora, Legora

Cash on delivery
As products assume more of the work, the unit customers pay for can change. Sierra makes this explicit through outcome based pricing for its customer service agents, with agreed criteria for results such as resolved support enquiries.

It also recognises that some interactions suit consumption based or blended pricing. The broader implication is that revenue can become more closely tied to successful execution. Defining completion, deciding when escalation is necessary and establishing evidence of success become part of both the experience and the commercial agreement. Sierra

This creates an important tension. The customer may pay for an outcome, while the provider incurs costs throughout the attempt to produce it. Models, computation, external tools, retries and human intervention all contribute to delivery cost. Software has always had operating costs, but agentic execution makes the amount of work required for each result especially important. Growth in usage or revenue does not, by itself, establish an attractive business. Leaders need to understand the cost of a successfully completed task, including failures and exceptions, and how that cost changes as the product takes on more demanding work.

AX decisions therefore affect margins directly. Clearer instructions can reduce wasted execution. Better context can prevent avoidable errors. An approval checkpoint can stop an expensive mistake, while excessive checkpoints can erase the benefit of delegation. Model selection, permissions and human review belong in the same economic assessment. The objective is to deliver a valuable result at a cost the customer and provider can sustain.

The intelligence moat
When companies acquire model capability from external providers, they must determine what they themselves contribute that customers cannot easily obtain elsewhere. That value can sit in proprietary context, specialised workflows, integrations and the reliability with which the product completes work. Dust illustrates this combination by bringing company knowledge, connected tools and multiple models together with permissions and activity monitoring. Its distinction between what an agent can access and who can use it makes authority part of the product’s working architecture. Dust

These capabilities have an economic purpose. Relevant context reduces the effort needed to explain a task. Reliable integrations allow the system to execute it. Permissions and monitoring make it possible for an organisation to authorise consequential work. Together, they can increase the scope of work a customer is willing to delegate and reduce the cost of delivering it. Governance consequently becomes part of productive capability: it establishes the conditions under which the system can be trusted with more responsibility.

For product leaders, this changes how architecture should be evaluated. A model improvement matters when it produces better results, lowers delivery costs or enables useful new responsibilities. A connection to another system matters when it removes a handover or makes a task possible. Technical investment should be tied to the work the product promises to complete. That is how available intelligence becomes a differentiated business.

Outcomes over process
The same reasoning applies inside the organisation. Many existing workflows reflect the limitations of earlier software: employees transfer information, reconcile records and coordinate steps across disconnected systems. Automating those activities can improve productivity, but it can also preserve work whose original justification is disappearing. If several handovers exist because systems cannot share context, leaders should examine whether those handovers are still needed before investing in making each one faster.

The opportunity is to reorganise around the outcome, retaining people where their judgement contributes most and delegating execution where the system performs reliably. Chatbase offers a modest but concrete example of how a focused AI application company can assemble external capabilities around its own product.

Stripe reports that Chatbase reached $10 million ARR in March 2026 with 26 employees, and that external billing infrastructure helped it avoid creating a dedicated billing team. This does not demonstrate an autonomous company. It shows how choices about what to own and what to acquire can support revenue scale without reproducing every function internally. Stripe’s Chatbase case study

Established software businesses must also examine whether their commercial incentives fit this new division of labour. If revenue depends on the number of people operating a product, reducing the need for those operators can create a conflict. If the business charges for outcomes, it must be able to measure and deliver them economically. Product strategy, pricing, staffing and operational accountability need to evolve together. Adding agentic capability without addressing these relationships leaves the underlying business only partly adapted.

Re/designing business
The economic significance of the AX Shift is that software companies can assume a greater share of the work for which customers currently supply the labour. That expands what they can sell, but also what they must deliver. Value depends on understanding the customers’ needs and the effort removed; profitability depends on the cost of reliable execution, competitive advantage depends on capabilities that make the product worth trusting with the task. These are the connected consequences of the same transfer of responsibility.

Product leaders must decide which work their products will take responsibility for and re/design their businesses to fulfil that commitment. The Big Boom supplies more intelligence and more software. Turning that abundance into a durable business requires a coherent relationship between customer value, delegated authority and delivery economics. That is the leadership challenge of the AX Shift.