". . moving fast is not the same as knowing where you're going . ."
Over the last few months I've been having a version of the same conversation with senior leaders across a range of organisations, and the pattern is consistent enough now that it feels worth a wider conversation.
It usually starts with AI. Specifically, with the fact that budgets have been approved, mandates have come down, and significant investment is already in motion. I had one conversation recently where a senior leader told me they needed to hire an AI Architect. When I asked what the person would be building, there was a pause. They didn't know yet. And when you get beneath the surface and ask some straightforward questions about what problem is actually being solved, what the data estate looks like, how any of this connects to the operating model, the answers tend to get uncomfortable quite quickly.
The pressure to move on AI is real, it's coming from boards, and it's not going away. But there's a difference between moving fast and moving with a clear sense of where you're actually going, and right now I'm seeing a lot of the former.
It goes deeper than the tools
What concerns me isn't just that some organisations are buying the wrong technology, though that is happening. It's that the whole approach is being compressed in ways that are storing up significant problems.
Strategy is being substituted by pilots. Data governance conversations are being deferred because they're slow and they don't make for a compelling announcement. The operating model implications of AI aren't being worked through properly because that requires a harder and more honest conversation with the business than most organisations are ready to have right now. And so you end up with AI capability that sits on top of fragmented data, disconnected from any clear outcome, and owned by nobody in a way that will actually hold up under scrutiny.
Someone I spoke to recently described their organisation's position as having the budget, having the tools, and not really knowing what they were doing with either. It was said with a degree of humour, but I don't think it was entirely a joke, and I suspect more people would recognise that description than would admit to it publicly.
A few things I keep coming back to
The framing of AI as primarily a cost reduction exercise worries me. Not because reducing cost is wrong, but because it tends to mean that the harder questions about process, data quality and operating model design get skipped in favour of an answer that looks clean on a slide. AI doesn't fix a broken process. It tends to make a broken process faster and more difficult to argue with.
The training programmes I'm hearing about are, in many cases, being delivered by vendors and focused on their own products rather than on how AI actually changes the way people work, make decisions, or are held accountable. Those are very different conversations and I don't think enough organisations are having the second one.
And the question of accountability is genuinely blurry in a lot of places. When an AI implementation doesn't deliver what was expected, working out who owns that outcome is harder than it should be, and that ambiguity was baked in from the start.
What I'm seeing ahead
My honest expectation is that the next twelve months will surface a lot of this. The gap between what was committed to and what has actually been built is going to become harder to manage, and closing it is going to require proper architectural thinking, clear problem definition, and in some cases a willingness to stop and reset before going further.
That said, I'm seeing a particular cross-section of the market and I know my view is shaped by that. There are organisations moving quickly and doing it well, and I'd genuinely like to understand what's making the difference for them.
So if you're in the middle of this, what's your experience? What's working, what isn't, and what do you wish had been tackled differently at the start? I'd welcome the pushback as much as the agreement.
Entasis Partners is a strategy and architecture solutions provider. We help organisations build the architecture capability that makes AI, data and transformation hold up, through consultancy, as-a-service delivery and building the teams to sustain it. If this resonates, book some time with me.
This article first appeared on LinkedIn, where you can join the conversation.
