Find the use case worth building, before anyone writes code.
Most AI budgets are spent on the first idea someone had in a meeting. We map the jobs your business needs done, weigh each against what it would take and what it would pay, and leave you with a ranked list and one pilot you can defend to your board.
What it covers
Jobs to be done interviews
We sit with the people who do the work today and ask what they did the week before, not what they would like. The job that keeps coming up is the one worth automating.
Value against complexity
Every candidate is scored on what it saves or earns and on how hard it is to get right in production. The first pilot is the one most likely to pay, not the most impressive.
A pilot scope in writing
One narrow question, the data it needs, how success will be judged, and what stays out of version one. Written down before any build starts.
The honest no
Sometimes the answer is that the data is not there yet, or the job is better done with a rule than a model. We say so, and you have saved a quarter.
How we do it
Two weeks, on site or on calls
Discovery sessions with operations, finance and whoever owns the outcome. We are looking for the moment work stalls, not for places to put AI.
A user story map of the current job
Every step, old and new, drawn out so the first release is a thin slice of the whole and nobody discovers a missing step in month three.
Feasibility on your own data
A day or two testing whether your documents, recordings or records can actually answer the question, before anyone commits to a build.
A ranked list and a first pilot
You get the candidates, the scores, the reasoning, and a pilot scope with a price. Take it to us or to anyone else.
- 1
Two weeks, on site or on calls
Discovery sessions with operations, finance and whoever owns the outcome. We are looking for the moment work stalls, not for places to put AI.
- 2
A user story map of the current job
Every step, old and new, drawn out so the first release is a thin slice of the whole and nobody discovers a missing step in month three.
- 3
Feasibility on your own data
A day or two testing whether your documents, recordings or records can actually answer the question, before anyone commits to a build.
- 4
A ranked list and a first pilot
You get the candidates, the scores, the reasoning, and a pilot scope with a price. Take it to us or to anyone else.
What you get
- A ranked list of AI candidates with value and feasibility scores
- A user story map of the job the first pilot replaces
- A written pilot scope, including what stays out of version one
- A feasibility note on your data, with what is missing
- A fixed price for the pilot
Where we have done it
Talent Carriage started with discovery sessions and a user story map before any build. The first version was still thrown away and rebuilt, which is exactly what the map made cheap.
Talent CarriageIndia Avenue asked for an assistant that emails documents. Discovery found the portal had no API, so that feature stayed out of version one, with the reason written down.
India Avenue
Questions we get asked
How long does discovery take?
Two weeks to start. That is enough to map the job, test your data and scope the first pilot, and it is deliberately short: you are not paying for a long study before anything gets built. After that, discovery does not stop. It runs alongside delivery, because the things we learn about your domain in month two change what is worth building in month three. Continuous discovery is how the pilot stays pointed at the problem that matters.
Do we have to build with you afterwards?
What if you find nothing worth building?
Who from our side needs to be involved?
Also under ai engineering
Talk it through
Tell us what you need from ai discovery and value mapping.
Thirty minutes with one of our architects. We will tell you whether it is a pilot, a build, or not worth doing yet.