For the last three years, the meaning of AI transformation has kept moving. After all that work it is tempting to feel we are well into the game, but the truth is we are still in the early innings, and the biggest shift is only now starting to gather pace.

The first act started in 2024 with co-pilots: 10 to 20 per cent efficiency in the functions that adopted them, plus a thin layer of revenue uplift. 2025 was a step up: leaders rebuilt core workflows from the ground up, with agents at the centre and people moved up to set direction, provide oversight and handle the edge cases - with the first truly agentic products launched.

In 2026 it has gone deeper again. In the best engineering teams across our portfolio, humans now write almost none of the code and productivity per developer is up fivefold. In support, agents resolve around 70 per cent of cases, now across most of the work rather than the narrow slice it used to be. Onboarding, customer operations and go-to-market are all heading the same way. That is real progress, hard won, and yet it is still the smaller part of what is coming.

Operating model and business model, rebuilt in parallel.

What makes 2026 different is that this is no longer only an operating model story. The agentic platform shift has become a business model shift too. The agentic products our companies launched last year are starting to land with customers, and they bring demands we have not had to handle before: different pricing and commercial models, new adoption dynamics, different challenges for the back office, and a different focus from leadership.

Twenty years ago, the move from on-premise software to SaaS forced B2B tech companies to rebuild how they priced, sold, supported and accounted for what they made. Many enterprise software vendors did not survive the leap. The agentic shift is asking the same question now, and on a faster clock: early data suggests it is moving several times quicker than SaaS did.¹ Companies will not have a decade to get it right.

The shift is already showing up in the numbers: across our most advanced portfolio companies, AI products built in the last year are adding more than 10 per cent to bookings, and several are above 25 per cent.

A new type of challenge

So we now have two transformations running at once, and together they add up to the deepest organisational change most companies will go through in a decade. Existing processes have to be reimagined from first principles, new operating models built for products that did not exist a year ago, yet existing software services maintained - and all of it under technology that keeps moving, so nothing stays settled for long. It is happening, too, while people are anxious about their jobs and, in some cases, learning to manage agents rather than other people.

The immediate answers focus on AI engineers: the market is noisy with the war for AI talent, and the logic feels sound, because a technology shift surely needs technologists. What we have learned at Hg together with our portfolio working through dozens of portfolio transformations, is that technical expertise is essential but only a fraction of what it takes. Rewiring a business this deeply takes five capabilities around the leadership table:

1. Technical depth: finding sharp AI talent able to access the full potential

Better AI models raise the floor, but they raise the ceiling faster still. With progress this quick, you have to read where capability is heading, and build for that. That means real AI expertise on the team, and a live connection into the frontier labs and the wider Silicon Valley flow. Without it, you end up building for yesterday's model and missing what is about to become possible. The real experts command high prices, and it is a false economy to compromise here. Picking them is hard, too, because track records in this field are naturally short, so be prepared to do the leg work.

2. Strategic vision: reimagining from first principles

The prize here is reimagination, not optimisation. The biggest moves come from going back to first principles: what value could we really create for customers, and how would we deliver our core processes if we were starting today? ‘Incrementalism’ is the enemy. The team asking how to make something 20 per cent better will lose to the team asking that bigger question.

3. Functional expertise: the rise of the OpEx role

Theory does not get this done: you need people who have actually run the function and know what a solution has to deliver in practice, how to make it work inside a real organisation, how to carry the team through the change, and how it connects to everything around it. This is where a new functional ‘OpEx’ role is emerging: an embedded builder, a 10xer, sitting alongside each functional leader and continuously rebuilding the work around agents as the technology and the business move on. These have a range of different titles – for example ‘GTM engineers’. The best teams embed AI talent with operational leaders across all functions, working hand in glove, ensuring fast iteration cycles. This aligns well with how successful agentic products are being built, with domain experts a core part of the product org.

4. Change leadership: more than just programme management

This is much more than a PMO exercise. You are reimagining how the company works while the ground keeps shifting under you, and asking people to do it when many are uncertain about their own roles. Transformation resourcing is important, but a transformation office and a chief transformation officer cannot hold that on their own. The CEO and the CHRO have to be visibly out in front, leading with a deep understanding of the organisation and real empathy for what you are asking of people.

5. Financial and investment rigour: avoiding AI theatre

Without a firm hand from the CFO, AI gains get stuck inside functions or quietly handed back to customers through pricing, and the programme drifts into AI theatre: plenty of activity, lots of promising ‘leading efficiency metrics’, but little of it banked. This is not crude cost cutting but something more demanding: knowing when to leave the cost lever alone so adoption can build, when to put serious money behind an investment that could deliver a step change in growth or efficiency. It is a more long-term and, we’d argue, PE-minded way of holding the purse strings, balancing what you bank now against what you build to accelerate the business.

What can you do now? Take an honest inventory

Most leadership teams and even external partners will be strong in one or two of these but carry real gaps in the rest. The gaps matter because each one fails in its own particular way.

Miss the technical depth → you build for a model or tools that are already behind the curve.

Without real functional expertise → change looks good on a slide and then quietly fails to land internally, or with customers.

A team short on strategic vision → optimises around the edges when it should be reinventing the core.

Weak change leadership → leaves people stalled and uneasy, and the programme loses momentum.

If the financial discipline is not there → the value you create leaks out before it reaches the bottom line, or key initiatives get cut before they deliver.

Get your senior team in a room, be honest about where you really stand against the five, and decide what to do about the gaps: what to build up, where you borrow resource and expertise, and what you might even buy in.

What building Hg’s AI transformation engine has taught us

We have learned this the hard way building our own Value Creation team and delivering AI transformations across our Hg portfolio. Sharp AI talent is scarce, but it can at least be hired for. Far rarer, and far more decisive in whether a transformation landed, were the people who paired a practical operator understanding and empathy with genuine AI fluency in the same head. The second lesson was that technical depth on its own stalls: without a well-designed transformation plan, functional and domain experts directly involved, a CEO and CHRO willing to lead the change, and the commitment to KPIs and forcing financial rigour.

That is why we have built the team the way we have: AI specialists working alongside functional experts who have run the function for real, transformation leaders embedded with our portfolio CEOs and their CHROs, and a long-term PE discipline on investment and returns running through all of it. It is intensive, and an ambitious transformation in a single Hg company can pull in 10-20 Hg team members, and be the primary focus for the company’s leadership team. But where the five come together the results are clear: our leading companies are running engineering at +5x times the old pace, and new AI products are adding more than 25 percent to bookings less than 12 months in.

Focus on the gradient, not the intercept

A lot of the narrative focuses on whether a business has an AI advantage and a headstart. But the race ahead is long and the change it demands is deep. We'd suggest that bringing in the right transformation expertise and resources deserves just as much attention, not least because it is so hard to do well.

¹ Deep Nishar and Nitin Nohria, "The End of One-Size-Fits-All Enterprise Software," HBR, April 2026. OPX AI