What the engagement actually produces
Most companies that come to us don't lack ideas about AI. They lack a way to tell a good idea from an expensive distraction. Every department has a wishlist — sales wants a lead-scoring tool, ops wants "something with AI" added to the backlog, someone saw a demo at a conference and can't stop thinking about it. None of that, on its own, is a strategy.
A strategy engagement turns that wishlist into three concrete things: a ranked list of where AI would genuinely move a number that matters, a decision for each item about whether to build custom, buy an existing tool, or leave it alone for now, and a written sequence for what happens in what order. It ends with a document your team can execute against, not a slide deck that gets filed away after the kickoff.
This sits deliberately upstream of build work. Building a custom agent or automating a workflow is far easier to scope well once you already know it's the right thing to build — working that out first is what this engagement is for. It sits alongside the rest of what we do; the full range of services shows what the build side looks like once a direction is set.
When you need strategy before you need a build
Not every project needs this step. If you already know exactly what you want built, roughly what it's worth, and what it needs to connect to, you can often go straight into scoping a build. Strategy work earns its cost when a few things are true:
- You have more candidate projects than budget or attention, with no reliable way to rank them against each other.
- The organisation is split on build vs buy, and the disagreement is slowing everyone down instead of getting resolved.
- An earlier AI initiative stalled or underperformed, and nobody is confident why — which makes the next attempt just as risky as the last one.
- Several teams want to move at once, and doing so without a shared sequence risks three groups building overlapping, conflicting tools.
If none of that describes your situation, say so on the first call. We'll tell you to skip ahead to scoping a build rather than sell you a strategy phase you don't need.
The opportunity audit
The audit starts with how the business actually runs, not with a list of AI use cases borrowed from somewhere else. We walk the workflows that eat the most hours or cost the most in errors, and score each candidate against a small set of questions: how often does this happen, how much genuine judgement does it require, what does it cost today, and what breaks if an automated version gets it wrong.
Most of what comes out of an honest audit doesn't look like AI at all. A workflow fed by bad data needs better data before it needs a model. A process that happens six times a month doesn't need automation — it needs a checklist. What survives that filtering is usually a shorter list than the one you walked in with, and a considerably more fundable one.
Build vs buy: the call that shapes everything after it
For each opportunity that survives the audit, the next question is whether to build something custom, buy an existing tool, or combine the two. This is the decision most companies get wrong by default — usually by going with whatever their most recent vendor conversation pushed them toward.
The honest version of this call weighs a few things against each other: how standard the workflow is (a standard workflow is usually well served by an off-the-shelf product), how deep the integration needs to reach into your own systems, how much of the logic is actually your competitive edge versus common practice, and what it costs to be wrong either way. A narrow, well-bounded task with a forgiving failure mode is often a buy. A workflow touching money, customers, or judgement calls specific to how you operate tends to justify going custom — which is where an agent or a purpose-built automation usually comes in.
We put this call in writing, with the reasoning behind it, so it still makes sense to whoever revisits it in six months.
Sequencing: what happens first, and why it's rarely the exciting one
Ranked opportunities don't get built in order of ambition. They get built in the order that teaches you the most for the least risk. The first project should be small enough to ship in weeks, close enough to the daily work that people notice it running, and low-stakes enough that a mistake is cheap to fix.
That first project also tells you things a planning document can't: whether the data is as clean as everyone assumed, whether the team will actually adopt a new tool, whether the integration is as straightforward as the API documentation implied. Everything after it gets sequenced with that evidence in hand, rather than on the strength of the original pitch.
The project furthest down the list is often the one that started the whole conversation — the ambitious one everyone wanted first. It usually belongs later, once the organisation has some practice working alongside AI rather than being the place that practice gets built the hard way.
Team enablement and workshops
A strategy that lives only in a document doesn't survive contact with a Monday morning. Enablement is the work of getting the people who'll actually use or manage a system comfortable with it before it shows up in their workflow.
In practice, that means working sessions with the teams affected — not a generic introduction to AI, but time spent on their actual workflow: what changes, what they're still responsible for, what the system won't do, and how to spot when it's got something wrong. We also work with whoever will own the system day to day, so someone on your side understands it well enough to make small decisions without calling us for every one.
The measure of a good workshop isn't attendance. It's whether the team can explain, a month later, what the system does and doesn't do — and whether they trust it enough to actually use it.
Governance and guardrails
Any AI system that touches real decisions needs boundaries agreed before it goes live, not written up afterwards as an incident report. Governance covers who can approve what the system does, what data it's allowed to see and retain, and who's accountable when it gets something wrong.
None of this needs to be heavyweight. A small business doesn't need a compliance department to do it properly — it needs a short, specific document: what a system may do on its own, what needs a human to sign off, what gets logged, and who reviews the log. We write that document as part of the engagement, in plain language, sized to the business that has to live with it day to day. For anything touching regulated data, we work within the general rules that already apply — data protection law and sector-specific rules like healthcare privacy regulation both impose real constraints — without standing in for your own legal counsel.
The honest recommendation is sometimes "don't build this"
A strategy engagement that always ends in "yes, build all of it" isn't doing its job properly. We've sat across the table from companies with a genuinely exciting idea and told them the timing was wrong, the underlying data wasn't ready, or the workflow needed fixing before it was worth automating at all.
That's not a failure of the engagement — it's the value of it. The cost of finding out a project was the wrong call during a strategy session is a few days of everyone's time. The cost of finding out after months of build is the whole budget, plus the team's trust in whatever AI project gets proposed next. We would rather say no early than build something that quietly never gets used.
How the engagement runs
Strategy work runs on the same structure as everything else we do: scoped and priced before it starts, and delivered the way we run every engagement.
- Discover. A free 30-minute call, then working sessions with the people closest to the workflows in question. We're mapping how work actually moves, not how the org chart says it should move.
- Audit and decide. We score the candidate opportunities, make the build-vs-buy call on each one, and write the reasoning down — including for the ones we recommend against.
- Sequence and enable. You get an ordered roadmap and the workshops that get your team ready to work alongside whatever comes first.
- Handoff or build. Some engagements end with a document your team executes internally. Others continue straight into a build, using everything the audit already established. Either is a good outcome — we don't need to be the ones who build it for the strategy work to have been worth doing.