, /ai-roadmap
We map your workflows, measure the hours they eat, and put a fixed price on the ones worth building. If nothing is worth building, we say so.
- →Every workflow inventoried and scored, with the evidence
- →Hours measured from your systems, not guessed in a workshop
- →The top candidates come back as signable proposals
- →The fee comes off your first build
Most AI advice stops one step before the useful part.
You can get a list of things AI could do for your business from almost anyone. A consultancy will sell you forty slides of them. A platform will let you generate your own. What nobody hands you is the number: which one first, how long it takes, what it costs, and when it pays for itself. Without that, the list sits in a shared drive and nothing gets built.
We give you the list with prices on it. Every recommendation we make is something we would build for a fixed price, on a fixed date. That constraint keeps the list honest, because we have to stand behind it.
Five stages.
One fixed price.
Inventory
We sit with each function head and list every recurring workflow: what happens, who touches it, how often, which systems. Usually twenty to forty.
Instrument
The part other assessments skip. Where the answer sits in your database, we query it instead of asking. Anything we could not measure is labelled an estimate.
Score
Six signals per workflow, each anchored to a written definition so you can check our work.
Cost
The survivors get a real build estimate. Fixed price, fixed date, a demo every two weeks. A number you could put in a contract, not a range.
Decide
Ninety minutes with your leadership. We walk the numbers and take the arguments. A good case changes the score in the room.
Six signals,
per workflow.
Every score is anchored to a written definition, published in the report. You can argue with any of them. In the decision meeting, people do.
A decision,
not a study.
The full report runs under thirty pages. There is no deck.
Opportunity register
Every workflow scored and ranked, with the evidence behind each score and a flag on whether we measured or estimated the number.
Data readiness assessment
Which systems can support automation today and which cannot, and what it would take to fix the rest.
Three costed proposals
Fixed price, fixed date, a demo every two weeks. Ready to sign.
Recoverable capacity
Hours per year, converted to money at your own rates, split by department.
The do-not-build list
Everything we considered and rejected, with the reason. A consultant who recommends everything is selling, not advising.
A two-page board memo
Someone who was in none of the meetings should be able to read it and approve a budget.
The fee comes
off the build.
Sign a build contract within sixty days of the presentation and the entire roadmap fee is credited against it. You pay for the thinking once, and only if you decide not to act on it.
Regulated sectors (financial services, global business, healthcare) carry a premium. They need a data-handling position we take seriously, and that adds work.
Built for companies between thirty and three hundred people.
Below thirty, the department diagnostic is usually the right size. Above three hundred, you probably have an internal transformation function already, and what you need from us is the build capacity rather than the roadmap.
The clients who get the most out of this share a profile: an ERP or accounting system with a few years of history, a lot of Excel, at least one process everyone complains about in every management meeting, and an owner who has been told to "do something about AI" with no way to judge which pitch on their desk is real.
Some things
we say no to.
Build a model when you have no labelled data to train it on.
Start with a chatbot because you asked for a chatbot by name. The diagnosis decides, not the request.
Run the engagement without access to your systems and your people. The measuring is the whole point.
Pad the list. If two workflows are worth automating and the rest of the budget should fix your ERP data, that is what the report says.
Questions,
answered.
Start with
one workflow.
Pick the process that annoys you most. Give us forty-five minutes and we will map it, measure what we can, and tell you roughly what it would cost to fix. No deck, no follow-up sequence.
If the number is interesting, we talk about the full roadmap. If it is not, you have lost forty-five minutes and gained a number you did not have.