Services
Seven kinds of work, grouped into the three moves they belong to. Most jobs are two or three of them at once, because the interesting problems sit where they meet.
Make the AI understand your business
One definition per number, one place it is computed, and a layer that carries the words your team uses. Skip it and an assistant pointed at your data will answer confidently and be wrong.
Getting your data in order
Pulling order, product, supplier and partner data out of the places it lives, cleaning it up, and keeping it current.
One rule underneath: every number is worked out in exactly one place, and everything else reads it from there. That is what stops a business having three answers to what it sold last month.
One definition for every number
A single layer that says what each number means and how it is worked out. Reports read from it. So do the AI assistants.
This is what makes AI on your own data work. It carries the words your team already uses, so a question asked in plain language lands on the right number instead of a plausible one.
Dashboards and reporting
Reports people act on, where margin means the same thing in finance, commercial and operations.
The hard part comes before the chart: agreeing what a word like "revenue" means when returns, rejections and partner fees all touch it, and deciding who owns that answer.
Work out what is worth doing
Once the numbers agree, the list of things worth building gets short. What is reachable this quarter, and what to drop.
Working out what to build
Which ideas are worth building, in what order, and which ones to drop. Usually a short piece of work before anything gets made.
These projects fail on the data underneath, not on the model. I look at what you already have, then tell you what is reachable in the next few months and what has to be fixed first.
Build it, and the ability to keep building
Pricing that sets itself, forecasts finance can plan on, reports the team runs without asking anyone. And your own developers made faster, so the next one does not need me.
Building the software
The thing itself: a system that sets prices, a forecast the business plans against, a tool your team opens every morning.
Where the data lives, the logic that makes the decision, the screens people click, and getting it running on your systems. At the end it works, with the notes to keep it working.
Connecting AI to your own systems
So someone can ask a question in plain language and get an answer out of your own order and product data.
Built on the definitions above, so the answers are right rather than merely fast. People reach only what they are allowed to, and nothing can change anything live by accident.
Making your team faster with AI
Your developers actually shipping quicker with AI coding tools, rather than just having them installed.
How to set a codebase up so the tools help instead of getting in the way, what to check before trusting what comes back, and where they should not be used at all. I work this way on my own products every day.
How engagements work
Bring the job that is slow, done by hand, or where nobody trusts the numbers, and you get a straight answer on whether software fixes it, including when it does not.
martin@enumstudio.com