An organisation is built for AI when its operating model lets work be redesigned quickly: clear ownership of outcomes, decisions pushed to where the information is, and governance that approves change in weeks, not quarters. Most hierarchies were designed for stability, so boards should redesign the operating model deliberately rather than wait for AI to strain it.
This question is now being asked openly by AI investors and technology leaders: can companies adapt as fast as the technology? This article gives the AIPEX view on what that means for the way a business is organised and governed.
Why does AI put pressure on the operating model?
AI changes the unit economics of work faster than organisations change their structure. A task that took a team a week can take an afternoon. A report that passed through four layers of review can be drafted, checked and summarised in minutes. When that happens, the bottleneck moves from doing the work to deciding what to do with it, and most organisations have not redesigned how decisions are made.
The evidence says the redesign matters more than the tool. In McKinsey’s March 2025 global survey, the redesign of workflows had the biggest effect of any factor tested on an organisation’s ability to see EBIT impact from generative AI, yet only 21% of respondents using it said their organisations had fundamentally redesigned at least some workflows (McKinsey, The state of AI, March 2025). Buying capability is easy. Reorganising around it is where the value, and the difficulty, sits.
What does an AI-ready operating model look like?
It is not a new organisation chart. It is a set of design choices that let the business absorb AI safely and repeatedly. We look for six:
| Design choice | Traditional model | AI-ready model |
|---|---|---|
| Ownership | Functions own activities | Named executives own outcomes end to end |
| Decision rights | Escalate up the hierarchy | Delegated to where the data sits, within set limits |
| Process | Automate the process as it is | Redesign the workflow around what AI makes possible |
| Governance | Annual approval cycles | Standing risk appetite and fast approval gates |
| Roles | Fixed job descriptions | Roles defined by judgement, oversight and exceptions |
| Measurement | Activity and headcount | Outcomes against a baseline set before the change |
None of these requires a restructure on day one. Each can be piloted in a single value stream, proved and then extended.
Where do operating models usually break first?
In three places. The first is ownership. AI initiatives that sit with IT or innovation teams often prove the technology and then stall, because no business owner has the authority to change the process. We describe the pattern in AI pilot purgatory.
The second is middle management. When AI compresses the work that layers of review used to do, managers can feel their role disappearing. Unless their role is redefined around judgement, exceptions and coaching, they become the quiet brake on adoption.
The third is governance speed. A risk committee that meets quarterly cannot govern tools that change monthly. The answer is not less governance. It is a standing risk appetite the board has approved in advance, with clear thresholds for what management may approve alone and what must come back.
How fast should the board move?
Faster than the budget cycle, slower than the hype. We recommend boards treat the operating model as a programme with phase gates: diagnose where AI would change the economics of work, prove a redesigned workflow in one area, industrialise the controls and data it needs, then scale. No gate is passed without evidence.
The risk of moving too slowly is real. Competitors that redesign first will deliver more per person and decide faster. But the risk of moving without design is also real: tools layered onto an unchanged structure produce cost without return.
The board should also look at its own part in the operating model. If every significant AI decision has to reach the board, the organisation will move at the pace of the board calendar. If the board sees nothing until something goes wrong, it cannot oversee the risk. The better pattern is a clear division: the board sets direction, risk appetite and the measures it wants to see, and management reports against them on a regular cycle. That gives management room to redesign work quickly while the board keeps sight of whether the changes are delivering and whether the risks remain inside agreed limits.
What should the board ask management?
- Which three workflows would change most if AI did the routine work, and who owns each end to end?
- Have we redesigned any workflow around AI, or only automated the old one?
- What decisions could be made lower down if people had better information, and what limits would apply?
- How quickly can we approve, launch and stop an AI change today, and what slows it down?
- How are we redefining management roles as routine review work shrinks?
- What baseline will we use to show the new model is working?
Where to start
Pick one value stream and redesign it properly, rather than spreading tools thinly across many. Our briefing on how to scale AI across the business, from pilot to production sets out the six conditions for scale, four phase gates and a 100-day roadmap. The step on the AIPEX roadmap that fits most boards here is Step 3, Plan: the Executive AI Strategy Day (£1,499 + VAT) agrees priorities, named owners, a benefits scorecard and a 90-day plan with your leadership team in one day.
Frequently asked questions
What is an AI-ready operating model?
One that lets work be redesigned quickly and safely: named owners for outcomes, decision rights delegated to where the data sits, a standing risk appetite with fast approval gates, roles built around judgement and oversight, and results measured against a baseline set before the change.
Why do AI programmes stall after the pilot?
Usually because no business owner has the authority to change the process, so the technology is proved but the workflow stays the same. Middle management and slow governance cycles are the other common brakes. See AI pilot purgatory →
Does becoming AI-ready mean a restructure?
Not at first. Each design choice can be piloted in one value stream, proved with evidence and then extended. Boards should treat the operating model as a phased programme, passing each gate only when there is evidence the redesigned workflow is delivering.
Correct as of 28 September 2026. This content reflects the law, official guidance and deadlines as published on that date. The EU AI Act and its guidance, standards and deadlines are still being amended and clarified. If you are reading this after 28 September 2026, check for later amendments, European Commission guidance and harmonised standards before relying on it. It is general information, not legal advice, and AIPEX accepts no liability for decisions taken on it without specific professional advice.