What’s really blocking AI progress? Insights from 200 IT leaders
AI investment is accelerating – but genuine progress is still a challenge.
What’s blocking AI progress
Over the past six months, Version 1 has hosted more than 200 senior IT and technology leaders across a series of closed-door forums and roundtable events, bringing together CTOs, CIOs, Heads of Technology, and senior programme leaders from commercial organisations across the UK and Ireland. These were senior operators with budgets, boards to answer to, and live AI programmes to manage, and the conversations reflected that.
What to expect from this report and who it’s for
Each finding comes with our analysis of the pattern, what the organisations getting it right are doing differently, and what it means in practice. Each finding also closes with a concrete recommendation you can act on immediately.
This report is based on conversations with senior technology leaders from commercial organisations across the UK and Ireland, the majority operating in
regulated sectors. The findings are specific to the private sector, where the investment mandates, governance pressures, and delivery challenges differ materially from public sector contexts.
FINDING 1: Governance has moved. Has your AI programme?
Eighteen months ago, the honest answer for any leader operating in a commercially sensitive or regulated environment was that the tooling for responsible enterprise AI deployment was still maturing, security frameworks were incomplete, model behaviour was harder to audit at scale.
The responsible default was to run controlled pilots while waiting for the governance landscape to settle into something more workable.
That has changed materially, and the leaders we spoke to who are operating in these environments are no longer treating governance as a reason to slow down. They are treating it as a capability to build, and the distinction matters because it changes where the investment goes and what the programme team spends its time on.
What we see working consistently is a technology-agnostic centre of excellence: a governing function that defines value at the process level, sets and enforces standards across the organisation, and retains control of outcomes regardless of which vendor’s technology sits underneath it. This does not require a large central team; several of the most effective examples we heard about were running lean functions with the ability to scale capacity in and out as specific initiatives demanded it, rather than trying to staff for peak demand across the whole portfolio.
“The organisations still treating governance as an obstacle tend to be the ones that have not yet built this central function. The ones that have built it find that governance becomes a competitive advantage, not a constraint.”
What has replaced governance as the primary blocker is something more structural: the absence of a clear, working decision framework for what to build internally, what to buy from the market, and what to bring in as an externally managed capability. The organisations still stalled at the planning stage are, in the majority of cases, the ones that have not yet resolved this question, and without resolving it, every new initiative triggers the same debate from scratch.
RECOMMENDATIONS
Build governance as a capability, not a control gate.
Establish a small, technology-agnostic centre of excellence with clear authority over AI standards and outcomes measurement. Define your build-buy-partner framework explicitly so that every new initiative has a decision path rather than triggering a debate. If you are in a regulated environment, design your QA framework now, in parallel with your next delivery cycle, before you need it.