MIT's AI Framework Is Right. Just Not for You.

MIT Sloan published a framework for AI transformation this week. It is serious work. It is also, almost certainly, not designed for you.

That is not a criticism of MIT. It is a clarification about who their research is built for. And understanding the gap - between what enterprise AI frameworks assume and what UK mid-market leaders actually have - is the most useful thing I can offer you right now.

What MIT actually said

Let me unpack the MIT framework for you, because the coverage has been thin. MIT Sloan's researchers argue that organisations should treat AI as an "operating system" - not a toolkit you layer on top of existing processes, but a foundational layer that redesigns how work, workforce, and workplace decisions interact. Their framework centres on four things: identify specific business problems worth solving, involve the people who do the work in designing the solution, test at small scale before going wide, and build governance and trust from the very beginning.

That is directionally right. I think it is actually the right framing. The AI Operating System metaphor is useful precisely because it insists AI has to change the structure of how decisions get made, not just speed up individual tasks. MIT's separate maturity research tracks how organisations move from isolated pilots through to enterprise-wide transformation - and the journey they describe is real. Most companies are stuck in the early stages. The World Economic Forum's 2026 analysis of organisational AI transformation says the same thing from a different angle: the organisations making progress are the ones redesigning how they work, not just adding tools.

The problem is what MIT assumes you already have when you start this journey.

The assumptions buried in the framework

Enterprise AI transformation research - MIT, BCG, WEF, all of it - is built on certain givens. Cross-functional C-suite governance teams. Dedicated data infrastructure. A CDO or CTO with a team. The ability to run parallel workstreams. The budget to engage specialist implementation partners. The organisational slack to run pilots that don't directly contribute to this quarter's numbers.

These are not unreasonable assumptions for a Fortune 500 company. They are wildly unreasonable assumptions for a 200-person UK professional services firm, a 500-person manufacturer in the Midlands, or a regional retail group navigating margin compression.

Vanguard's 2026 analysis of AI's economic impact makes the point bluntly: AI-driven productivity gains are concentrated in firms with pre-existing digital infrastructure. Smaller organisations without that base face structural barriers that enterprise frameworks simply don't address. What that means is that following the MIT playbook without adjusting for your constraints doesn't just slow you down - it points you in the wrong direction. You end up trying to build infrastructure you don't have the resources to build, rather than getting value from what you already have.

How do UK mid-market businesses adopt AI practically?

UK mid-market leaders don't need a scaled-down Fortune 500 playbook - they need a different starting point entirely. Enterprise frameworks assume you're building AI into abundant resources. The practical alternative starts from scarcity: identify the two or three decisions or workflows that consume the most time relative to their strategic value, and redesign those first. You don't need a governance committee to do that. You need clarity about what matters and the authority to change it.

Three things mid-market can skip

This is where I want to be concrete, because the "adapt enterprise frameworks for your context" advice is usually useless without specifics.

First, you can skip the data infrastructure phase. MIT's framework assumes you need to build a clean, integrated data layer before you can do anything meaningful with AI. You don't. The tools available today - particularly the large language models that can reason across messy, unstructured information - work with what you have. Your team's emails, documents, meeting notes, and institutional knowledge are more useful than you think. Start there.

Second, you can skip the governance committee. Governance matters, but you don't need a cross-functional committee to get started. You need one person - probably you, or a direct report - who is accountable for how AI is being used and what the organisation is learning. Governance that takes six months to set up before you've done anything is governance that prevents learning.

Third, you can skip the consultant-led transformation programme. MIT's "last mile gap" research identifies the distance between AI deployment and measurable business impact as the central challenge. The organisations that close it do so by redesigning how work is structured, not by running large implementation projects. Structural redesign at your scale is something your leadership team can do. You know your business better than any external team will.

What is the AI operating system framework for smaller businesses?

For UK mid-market organisations, an AI operating system means building AI into how decisions, workflows, and roles are structured - starting from what you already have, not from what a larger organisation would build. In practice this means three things: picking the highest-leverage decisions in your business and asking how AI changes the information available to make them; identifying the workflows where your best people spend time on tasks that don't require their judgement; and creating a simple feedback loop so you know what is working. That is the whole framework.

The asymmetric advantage of being smaller

Here is the thing I think most enterprise AI research misses entirely. Smaller organisations have a genuine structural advantage in AI adoption. You can make a decision this week and see the result next week. You don't need to navigate procurement cycles, change management programmes, or organisational politics across multiple divisions. The "last mile gap" that MIT identifies - between deployment and impact - is shorter for you by design.

What that means is that you don't need to match enterprise resources to get enterprise-level returns from AI. You need clarity about where the leverage is in your business, and the willingness to change how those specific things work. The economics are genuinely asymmetric. Research from Accenture found that smaller organisations adopting AI in focused workflows can achieve returns comparable to enterprises ten times their size, because the distance between deployment and impact is simply shorter. A 200-person business that gets AI working well in its three or four highest-leverage workflows can compound that advantage faster than a 20,000-person organisation navigating the same transformation at scale.

There is a video I recorded recently that goes deeper on this - you can watch it here: https://youtu.be/qVoVkrk5mRU. And if you want the longer version of the framework and how to apply it to your specific context, this is exactly what the AI Leaders Fellowship is built around.

The MIT framework is right about what AI transformation requires. Structural change, not tool adoption. It is just that the structure you need to change is more accessible than they assume. That is not a disadvantage - it is precisely the opportunity.