When Execution Becomes Abundant, What Are Technology Leaders Actually Selling?.

AI is changing the economics of software development and technology consulting incredibly quickly. Code can be written faster, designs can be explored faster, requirements can be analysed faster, and tasks like testing, documentation and research can now happen in a fraction of the time they once took.

For software development agencies, technology consultants and Fractional CTOs, that creates an uncomfortable question: if AI means clients need less of our time to get things done, what are they actually paying us for?

It is a question that goes far beyond AI productivity. It gets to the heart of how technology services are valued, packaged and priced.

I recently came across Sequoia Capital’s article, Services: The New Software, which gave me a useful way of thinking about the answer. The article makes a distinction between intelligence and judgement. As Sequoia puts it, “Writing code is mostly intelligence. Knowing what to build next is judgement.”

That distinction matters. AI isn't necessarily going to kill technology consulting, software agencies or Fractional CTO services, but it is going to change what clients value and, therefore, what they should be paying for.

How AI is changing software development and technology consulting

For most of the history of software development, execution has been expensive. Building a digital product required significant amounts of specialist human time. Designers designed it, developers coded it, QA teams tested it, product teams wrote requirements, technical teams documented it and project teams coordinated everything.

Naturally, agencies and technology consultancies built commercial models around that constraint. We estimated how much human effort a project required and sold that effort, whether explicitly through day rates or indirectly through fixed project fees.

AI starts to challenge that model. If an experienced developer working with AI can achieve something significantly faster than they could previously, the value of the resulting software hasn't suddenly decreased. The cost and effort involved in producing it have changed, but those are not the same thing as value.

This is why technology businesses need to be careful about equating speed with value. If AI allows a team to solve a £500,000 business problem in four weeks rather than four months, it doesn't follow that the solution has become less valuable. In many respects, the opposite is true because the client receives the outcome sooner.

We have written more about this shift in How AI Is Changing How Software Gets Built, including where AI can accelerate delivery and where experienced engineering judgement still matters.

Why judgement is becoming the differentiator in technology services

There is another important consequence of faster execution: knowing what to execute becomes considerably more important.

AI can generate 20 potential features incredibly quickly, but it cannot automatically tell you which three will actually change the economics of the business. It can generate an architecture, but someone still needs to understand whether that architecture makes sense given the organisation's scale, budget, legacy technology, security requirements and future plans. It can produce thousands of lines of code, but someone still needs to take responsibility for whether that code should exist in the first place.

That is judgement, and good judgement is difficult to automate because it is built from experience. It comes from having seen products succeed and fail, understanding users, knowing where projects typically go wrong and recognising when a seemingly simple requirement will create significant technical complexity later.

It also means knowing when technical debt is acceptable and when it is dangerous. Sometimes it means being willing to tell a client, “Don't build that.”

That advice might take five minutes to give, but it may represent 20 years of experience. Charging purely for those five minutes increasingly makes very little sense.

What are technology consulting clients actually paying for?

I think the answer is moving away from capacity and towards outcomes, judgement and accountability.

Historically, a client might effectively have bought 500 hours of design and development. Increasingly, what they should be buying is a team capable of getting them from a business problem to the right technology outcome.

That's a subtle but important shift because it changes the questions that matter. What should we build? What shouldn't we build? Where should AI be used, and where shouldn't it? Which assumptions should we test before investing heavily? What is the simplest version capable of proving the proposition? Which technical decisions will constrain us later? Where is the genuine commercial or technical risk? When should we change direction?

Those decisions can save substantially more money than simply producing code faster.

That is also why good digital product development increasingly starts before a line of production code is written. Product strategy, validation, design and technical decisions all reduce the risk of executing the wrong idea efficiently.

How AI is changing the Fractional CTO model

Fractional CTOs are perhaps one of the clearest examples of this shift.

A business shouldn't really be buying eight hours of a CTO's time every Friday.

It is buying access to senior technology judgement without employing a full-time CTO.

That might include architecture decisions, technical due diligence, supplier management, technology strategy, AI adoption, product direction, security, recruitment or simply having an experienced person available when an important decision needs to be made.

AI should increase the leverage of that person.

If research, documentation, analysis and technical investigation can happen faster, a Fractional CTO can spend more time on the areas where experience matters most.

The commercial model therefore becomes less about: "How many days do you need me?", and more about: "What technology responsibility am I taking ownership of?"

That is a much more interesting proposition.

For businesses buying Fractional CTO services, the value is not simply access to a set number of hours. It is access to senior technology leadership, better decision-making and accountability at points where the wrong choice can become expensive later.

What AI means for software development agencies

The agency model is going through a similar transition.

For years, software development agencies have sold teams.

Two developers. One designer. Half a project manager. Six months.

AI-assisted software development should start changing that conversation.

Clients shouldn't necessarily care how many developers were required.

They care whether the product works.

Whether users adopt it.

Whether it solves the business problem.

Whether it is secure and scalable.

Whether it launches on time.

Whether the investment generates a return.

And whether the people making decisions along the way know what they're doing.

The agency of the future therefore isn't simply a more efficient software factory.

It is a technology partner with an increasingly powerful execution engine underneath it.

That distinction is important.

AI can make software delivery faster, but speed alone is not the product.

The value sits in combining that speed with product thinking, technical judgement, commercial understanding and responsibility for the outcome.

That is the thinking behind Arch's approach to AI software development solutions: using AI to increase the capability of experienced product and engineering teams rather than treating automation as the outcome itself.

How should technology consulting services be packaged?

This potentially changes how technology consulting and software development services are packaged.

There will still be situations where day rates and time-and-materials models make sense.

But I think we'll increasingly see services organised around responsibility and outcomes rather than capacity.

Discovery becomes less about selling workshops and more about reducing investment risk.

Prototyping becomes about answering important questions before committing significant capital.

Fractional technology leadership becomes an ongoing strategic capability rather than a number of days per month.

Software development becomes responsibility for delivering a defined outcome rather than simply supplying engineering hours.

AI implementation becomes identifying where automation creates genuine commercial value rather than adding AI because everyone thinks they need it.

The deliverable isn't always the work.

Sometimes the deliverable is certainty.

That is particularly relevant to AI-assisted product development, where the ability to move faster only matters if you're moving in the right direction. Arch's AI-accelerated MVP and product development approach is built around validating the proposition, technical direction and product before committing to the full build.

How should AI-era technology services be priced?

This is probably the more difficult question.

If AI makes an experienced team twice as productive, simply halving the price doesn't create a sustainable model.

It also misunderstands what the client is purchasing.

Imagine two technology teams.

One takes six months and £200,000 to solve a problem.

Another uses better technology, AI and 20 years of experience to solve the same problem in three months for £150,000.

The second team has used fewer hours.

But it has arguably delivered considerably more value.

Pricing technology purely according to the hours required therefore becomes increasingly disconnected from the outcome.

I think we'll see more combinations of fixed pricing, retainers, productised services, milestone-based engagements and outcome-oriented commercial models.

Not because time suddenly becomes irrelevant.

But because time becomes a worse proxy for value.

That shift matters for clients as much as it does for suppliers.

A pricing model built around outcomes creates a different conversation. Instead of focusing primarily on effort, it puts more emphasis on what success looks like, where risk sits and what the technology is actually expected to achieve.

Will AI replace technology consultants and Fractional CTOs?

There is an uncomfortable side to this.

Sequoia argues that today's judgement can become tomorrow's intelligence as AI systems learn from increasingly large amounts of domain-specific data.

So simply saying "AI does the execution and humans provide the judgement" isn't a permanent business model.

The boundary will keep moving.

Tasks we currently consider highly skilled will become increasingly automated.

That means people working in technology need to keep moving up the value chain.

From writing code to designing systems.

From designing systems to understanding businesses.

From understanding requirements to challenging them.

From delivering features to identifying opportunities.

From supplying expertise to taking responsibility for outcomes.

The safest place isn't simply being good at using today's tools.

It is being the person clients trust to decide which tools to use, what problems to solve and what decisions to make next.

The future of technology consulting in an AI-first world

I don't think AI means the end of agencies, consultants or Fractional CTOs.

I think it forces us to become clearer about why we exist.

If our value is simply producing outputs that AI can increasingly produce, then our value will inevitably be squeezed.

But if our value is understanding the problem, applying experience, making difficult trade-offs, reducing risk and taking responsibility for the outcome, AI can make that expertise considerably more powerful.

Execution is becoming cheaper.

Judgement isn't.

And perhaps the biggest opportunity created by AI isn't simply doing the same work faster.

It's finally moving technology services away from selling time, and towards selling what clients actually needed from us all along:

Better decisions and better outcomes.

Frequently asked questions

How is AI changing technology consulting?

AI is reducing the amount of human time required for tasks such as research, documentation, software development, testing and analysis. As execution becomes faster, the value of technology consulting increasingly shifts towards judgement, strategy, accountability and the ability to identify which problems are worth solving.

Will AI replace technology consultants?

AI can automate an increasing amount of execution, but technology consulting involves more than producing outputs. Consultants still need to understand business context, assess risk, challenge assumptions, make trade-offs and take responsibility for recommendations. The balance between human and AI work will continue to change, but the need for strong technology judgement remains significant.

How is AI changing software development agencies?

AI allows software development agencies to accelerate activities such as coding, testing, prototyping and documentation. This creates an opportunity for agencies to move away from selling engineering capacity alone and towards taking greater responsibility for product outcomes, technology decisions and commercial impact.

What does a Fractional CTO do in an AI-driven business?

A Fractional CTO provides senior technology leadership without the organisation employing a full-time CTO. Their responsibilities can include technology strategy, architecture, AI adoption, technical due diligence, supplier management, security, recruitment and product direction. AI can increase their leverage by reducing the time spent on research, analysis and documentation.

What is outcome-based pricing in technology consulting?

Outcome-based pricing focuses the commercial relationship on the result being delivered rather than solely on the number of hours or days required. Depending on the engagement, this can include fixed project fees, retainers, milestone-based pricing or fees connected to defined business or technology outcomes.

Will AI make software development cheaper?

AI can reduce the cost and time required to execute some software development tasks. That does not necessarily mean the value of the resulting software decreases. If a business problem can be solved faster while maintaining quality, security and scalability, the organisation may receive value sooner.