From ‘we’re thinking about AI’ to live in a week

The gap between thinking and doing is where the real risk lives – and most organisations have been sitting in it for eighteen months.

Oli Brown, Territory Manager, Moterra 9 June 2026

“We’re thinking about AI” has become the new “we’re thinking about going digital.” I heard that phrase constantly in 2014. The organisations that were still thinking about it in 2016 found themselves two years behind the ones that had just started.

We’re in the same moment now. Except the gap is closing faster.

Most organisations I speak to have been “thinking about AI” for the better part of eighteen months. Committees formed. Consultants briefed. Vendor shortlists circulated. And yet – nothing in production. Nothing that’s actually changed how anyone does their job. Meanwhile, competitors are doing it. Quietly, without press releases, one workflow at a time.

The real risk isn’t moving too fast. It’s not moving at all.

The two-year strategy trap

The typical response to AI is to treat it like an ERP implementation. Form a steering committee. Hire a consultant to build a roadmap. Identify all the use cases across the entire organisation. Wait for the right platform, the right budget cycle, the right moment.

The perfect moment never comes.

AI is not an ERP. You don’t need to map every process, integrate every system, and get sign-off from every department before you start. You need one workflow that works. One thing that’s genuinely painful right now, made meaningfully better, inside a safe environment.

That’s the reframe. Not an AI strategy – a specific workflow that can be improved within a few weeks. Once that works, you do the next one. The strategy emerges from the doing, not the other way around.

The organisations I’ve seen get stuck are almost always the ones trying to solve everything at once. The ones getting value are the ones who picked something small and shipped it.

Three questions before you do anything else

Before you choose a tool. Before you talk to vendors. Before you book another workshop.

First: where is your team already using AI – officially or not? Because they are. If you haven’t done this audit, you’re making decisions without the most important data point you have. The answer tells you where the real demand is, and where the real risk already lives.

Second: does your AI actually see your business data – or just the internet? A generic AI tool gives generic answers. It doesn’t know your processes, your clients, your policies, your risk framework. It answers from its training data – which is not your business. That’s fine for writing a birthday card. It’s not fine for a client proposal, a compliance check, or anything that requires your organisation’s actual knowledge.

Third: do you control where your data goes? This one determines what you can safely use AI for. If the answer is “I’m not sure” – that’s the answer. And it’s the thing to fix first.

These aren’t philosophical questions. They’re diagnostic. The answers tell you exactly where to start – and exactly what’s already broken.

What ‘live in a week’ actually looks like

It’s not magic. It’s not a six-month implementation with a big reveal at the end.

The first step isn’t an audit. It’s a conversation.

Gather your team – informally, without an agenda that feels like an investigation. Ask them where they’re already using AI. Encourage honesty. If someone admits they’ve been using ChatGPT for client work, that’s not a disciplinary matter – it’s useful information. You want to know where the friction is. Where people are losing an hour a day to something repetitive. Where they’ve already found a workaround that happens to involve an AI tool they downloaded themselves.

That conversation – done without threat, done with genuine curiosity – will tell you more about where to start than any vendor evaluation or strategy document.

Step two is picking one use case. Not a strategy. One specific workflow that’s painful right now – a process that takes too long, produces inconsistent results, or relies on someone hunting through documents to find an answer. Something real, used by real people, that would be noticeably better with AI.

Step three is deploying on the right platform – securely, with your actual business data, in a controlled way. Then doing the next one.

Once you know where the friction is, the next question is: what kind of AI deployment actually fits your business?

There are two approaches worth knowing about.

The first is a purpose-built private AI platform – something like Moterra Business AI Suite. Fully private, running in your own cloud environment, customisable to your workflows and business data. It takes a little longer to set up than plugging in a SaaS tool, but you end up with AI that actually knows your business – your documents, your processes, your clients. And because it’s yours, your data never leaves your perimeter.

The second is a private deployment of tools your team already knows and trusts. Claude, built by Anthropic, is the leading example here. Most people know Claude as a public AI tool – but it can be deployed privately, inside your own infrastructure, so your team gets the familiar experience without the data exposure. This isn’t widely publicised yet – it’s emerged through the developer and thought-partner community rather than a formal product launch – but it’s real, and it works.

Both approaches solve the same core problem: AI that knows your business, in an environment you control. The difference is how much customisation you need and how fast you want to move.

That’s the iterative approach. Not a big-bang two-year strategy. One thing that works, then the next.

The window for this is open. It won’t stay open forever. Large enterprises are building private AI infrastructure right now – quietly, with hundreds of engineers, on proprietary data. Mid-size businesses have a window to move fast enough to matter. But speed is the advantage – and that window closes.

The one thing that separates the firms getting value from the ones still thinking

It’s not budget. It’s not technical expertise. It’s not even having the right tool.

It’s the decision to start with something real.

Not a pilot. Not a proof of concept that lives in a slide deck. A real workflow, with real data, used by real people, that makes someone’s job genuinely easier this week.

I lead the UK division within an AI company, so I have a stake in this. But that’s exactly why I see these patterns up close. The AI itself is not the hard part. Getting it to actually know your business – your documents, your processes, your clients – without handing all of that to a third party – that is.

The question isn’t whether AI will change how your business operates. It will. The question is whether you’re the one who controls how.

If you’ve been thinking about it for six months – that’s the answer

You’re already behind the curve. Not catastrophically. But enough that starting this week matters more than starting perfectly.

Pick one workflow. Ask those three questions. Deploy something real.

We’ll be at Trailblazing Tech 2026 on 18 June – if you want to see what both of these approaches look like in practice, come and find the Moterra team. We can walk you through real deployments – not demo videos, not slide decks. And if you leave with a clearer sense of which path fits your business and three things you can do on Monday morning, that’s a good outcome.