AI advisory
Most organisations arrive at AI with a solution looking for a problem — a board that wants a strategy, a competitor that announced something, a team that tried a chatbot and quietly stopped using it. We start from the opposite end: your actual work, and which parts of it a model can carry.
The pilot that never became anything
The common failure isn't technical. It's that nobody established, before building, whether the task was one AI is good at, whether the data to support it existed, or who would own the result once the pilot team moved on. So the demo works, the rollout stalls, and the organisation concludes AI doesn't work here — when what didn't work was the choice of problem.
You’ll recognise this if
- You have run a pilot that impressed people and changed nothing
- Different teams are buying different AI tools with no shared plan
- Someone has asked for an AI strategy and nobody knows where to start
- You suspect parts of your operation are automatable but can't say which
What we actually deliver
An opportunity map of your operation
We work through your workflows and mark each one: worth automating now, worth revisiting once the data is in order, or better left to people. The 'leave it alone' column is usually the most valuable one, and it is the column most vendors won't give you.
A costed shortlist
For each opportunity that survives, an estimate of build cost, running cost, and the ongoing effort to keep it accurate. Model inference is rarely the expensive part — the data work and the human review around it usually are, and we price those honestly.
A readiness assessment
An honest read on whether your data, systems, and team can support what you want to build. If the answer is not yet, you get the list of what has to change first rather than a project that fails six months in.
A build-or-buy recommendation
Much of what teams commission has already been built and can simply be bought. We'll say so when it's true, and tell you which parts are genuinely specific to you and worth building.
The way we approach it
Interviews before analysis
We spend time with the people doing the work, not only the people sponsoring the project. The gap between how a process is documented and how it actually runs is where most of the opportunity — and most of the risk — turns out to live.
Test the hard assumption early
Every plan rests on one assumption that would sink it if wrong: usually that a certain quality bar is reachable, or that a certain dataset is usable. We identify it and test it in days, before it is holding up a roadmap.
Written, and yours
The output is a document you own and can circulate — to your board, to another vendor, to a team you hire instead of us. It's designed to be useful whether or not you build with us.
What changes
- A ranked list of AI opportunities with cost and effort attached to each
- A clear statement of what your data and systems can and cannot support today
- A defensible answer when someone asks why you are — or are not — doing something
- A roadmap you own outright, useful with any vendor
Asked often enough to answer here
How long does an advisory engagement take?
Typically two to four weeks depending on how many workflows are in scope and how available your team is for interviews. We would rather run a focused two-week review of one department than a shallow pass across the whole company.
What if the honest answer is that we shouldn't use AI?
Then that is what the report says, along with what would have to change for the answer to become yes. We would rather deliver that finding than sell you a build we don't believe in — it is the cheapest useful thing we can tell you.
Do we have to build with you afterwards?
No. The roadmap is yours regardless, written so another team could pick it up and execute it. A good share of advisory clients do go on to build with us, but nothing in the engagement depends on it.
Related
Wherever you’re starting from, let’s figure out the next step.
Tell us what you’re building — we’ll tell you honestly whether we’re the right team for it.