Blog
AI Is Rewriting Consulting. GSIs Must Sell Outcomes
Andrew Doughty
Published on
Table of contents
Lorem ipsum
Lorem ipsum
Lorem ipsum
Lorem ipsum
Lorem ipsum
For decades, the economics of global technology consulting have been remarkably durable. Digital transformation created projects, projects required people, and more people meant more billable hours. That model built some of the largest and most successful technology services businesses in the world.
AI is beginning to challenge it.
This is not because enterprises will need less technology. The amount of software being created is likely to increase significantly. The challenge is that the human labour required to create it is falling, and that has serious implications for the consulting industry.
The end of headcount-led growth
The traditional model was straightforward: more transformation led to more projects, more people, more billable hours and more revenue.
AI introduces a different possibility. More transformation, delivered by AI agents and smaller expert teams, producing substantially greater output.
An AI-enabled engineering team can increasingly accomplish work that previously required several times as many people. Software, tests and documentation can be produced faster. Applications can be migrated faster. Incidents can be investigated faster. Agents will increasingly perform work that has traditionally sat inside large delivery teams.
This raises an uncomfortable question for the industry. What happens when the technology you are helping customers adopt also reduces the amount of labour they need to buy from you?
Markets are repricing the services model
The uncertainty is already visible in public markets. Investors are trying to determine whether AI represents the next major transformation opportunity for technology services businesses, or a structural challenge to an industry whose economics have historically depended on human labour.
On 22 September, TCS closed around 37% below its 52-week high and Wipro around 39% below its own. Accenture remains substantially down in 2026, even after its shares rose on the announcement of an AI safety partnership with Anthropic. IBM, with a materially different mix of software, infrastructure and services, has behaved differently.
The market is not necessarily betting against consultancies. It is asking what kind of consultancy will be valuable in an AI-native world. And beneath the short-term movements sits a more fundamental question: if AI allows technology projects to be delivered with far fewer people, what happens to businesses that have historically been rewarded for deploying more of them?
The consulting paradox: more demand, less effort
Two forces are at work simultaneously.
The first is compression. AI can automate increasingly sophisticated knowledge work. Coding, testing, analysis, migration, documentation and support are all becoming more productive. Work that once required hundreds of people may eventually require considerably fewer, which puts pressure on labour-based revenue.
The second is expansion. Enterprises now have a far larger range of things they could automate, modernise and reinvent. Legacy systems can be transformed, processes redesigned and AI agents deployed across organisations, with software created at a pace that was not previously possible.
The addressable opportunity could therefore grow considerably at the same time as the labour required to deliver each project declines. AI could create the most significant technology transformation opportunity in decades while undermining the traditional economics of the firms expected to deliver it.
The consultancies that succeed will be those that reconcile these two forces.
Why productivity alone won't protect the model
The obvious response is to use AI to make consultants more productive. Every firm will do this, and should. Strategically, however, it is not enough.
Consider a customer that previously paid for the equivalent of 100 people's effort to deliver an outcome. AI now allows that outcome to be delivered with 30. Who captures the productivity benefit?
Initially, the consultancy may, through higher margins. Over time, customers will ask why they are paying the same for significantly less human effort. That places pressure on any business model tied to people, utilisation, day rates and billable hours.
The more compelling opportunity is to change what customers are buying: moving away from people, hours and activity, and towards outcomes, IP, platforms, assurance and accountability. In practice, that means selling confidence that a release is safe rather than 50 testers. Selling a successful migration rather than the hundreds of engineers required to execute it. Selling continuous evidence that regulatory requirements are being met rather than the people needed to check them manually.
It means charging for the value and risk associated with an outcome, rather than the effort required to produce it. That is a fundamentally different business model.
Trust becomes the scarce resource
There is another consequence of this transition that I believe is underestimated.
Consider an enterprise software programme a few years from now. Instead of 500 developers and 200 testers, it might consist of 100 domain and engineering experts working alongside hundreds of autonomous agents.
Those agents could generate a very large volume of software. But someone still needs to answer the questions that matter. Did we build the right thing? Does it work? Does it comply with our requirements, policies and regulations? What has changed, and what risk does this release introduce? And can we prove all of it?
As the cost of producing software falls, the cost of being wrong does not. In banking, insurance, healthcare, government, telecommunications and other regulated industries, that cost can be severe.
The result is an inversion. AI makes production abundant. Trust becomes scarce. And what is scarce becomes valuable.
Touchstone and the case for agentic assurance
This shift is one of the ideas shaping the direction of Touchstone at Virtuoso QA.
We do not believe the long-term opportunity is simply to use AI to make traditional software testing faster. Testing itself is changing. The larger opportunity is to create an agentic assurance layer around increasingly autonomous software delivery.
Imagine an agent that understands the regulations an organisation must comply with, its internal policies and standards, its business processes and risks, its applications, workflows and business rules, its requirements and expected behaviour, and established testing and quality practice.
That context can be used continuously to determine what needs to be validated, generate and execute tests, maintain them as applications change, identify failures and produce attributable evidence of what has been checked.
The objective is not simply to answer "Did the tests pass?" It is to answer a more valuable question: "Should we trust this release?" That distinction becomes increasingly important as AI-generated software moves faster than people can reasonably review it.
From selling capacity to owning outcomes
This is why I do not see AI simply as a threat to the major systems integrators. For those prepared to change their model, it represents a significant opportunity.
Consider a consultancy telling a bank:
"We don't need 200 people testing your transformation programme in the traditional way. We will deploy an agentic assurance capability that continuously understands your requirements, regulations, applications and risks. Our specialists will govern the process, investigate exceptions and provide accountability. And every release will come with evidence showing why it can, or cannot, be trusted."
That is a very different proposition from selling testing capacity, and potentially a far more valuable one. The consultancy becomes responsible for an outcome, supported by technology, IP, agents and expertise, rather than supplying the labour required to perform the work.
The economics change accordingly. If technology allows a consultancy to deliver the same or a better outcome with a fraction of the people, that need not mean a fraction of the revenue. It could mean higher-value outcomes, stronger margins and much greater scale.
But only if the commercial model changes with the delivery model.
Where value moves next
Cloud moved value away from owning infrastructure. SaaS moved it away from installing software. AI will move it away from a lot of the repeatable knowledge work consultancies bill for today.
The integrators aren't going away. The strongest will combine industry knowledge, their own technology, AI agents and people willing to be accountable, and they'll deliver at a scale that wasn't affordable before.
The question every consultancy should be asking isn't how to add AI to its existing services. It's what clients will still pay for once AI has taken most of the effort out of those services.
In software delivery, we think a big part of the answer is trust. Moving fast, and being able to prove that what you shipped works, complies and is safe.
That's the bet we've made.
