AI data analysis tools
Interfaces that let non-technical people interrogate company data in plain language.
- Engagement
- Fixed-scope build, metric modelling included
- Typical timeline
- 8 weeks
- Starts with
- The twenty questions your team asks the data team most often.
Someone in commercial wants last quarter’s repeat-purchase rate by city. The analyst queue is four days long, so they export a spreadsheet and work it out themselves, arrive at a number that disagrees with the dashboard, and now there are two versions of the truth and a meeting to reconcile them.
Analysis in minutes without joining a queue for the data team.
What we build
Concrete artefacts, handed over and documented.
- 01
A semantic layer where each metric is defined once, what makes a customer active, when revenue is recognised, so every question resolves to the same definition
- 02
Plain-language questions translated into queries against read-only views, never against your production tables
- 03
The generated query shown next to the answer, so an analyst can check the logic and a sceptic can be satisfied
- 04
Charts chosen for the shape of the answer, with the underlying rows one click away
- 05
A library of verified questions, which is how the tool becomes more reliable over time rather than less
- 06
Row and cost limits on every query, so a careless question cannot slow the warehouse for everyone else
What changes
Routine questions get answered inside the meeting where they are asked
One definition per metric, so two people stop arriving with two numbers
The analyst queue holds genuine analysis instead of data extracts
How it runs
01Define the metrics
The unglamorous part, and the one that decides whether anybody trusts the tool: agreeing what each number means.
02Query generation
Question-to-query built against the modelled layer, scored on your twenty most common questions with known answers.
03Roll out to one team
One department uses it for a fortnight; every wrong answer becomes either a fix or a verified saved question.
Chosen per project. Named here so you can see the shape of it.
- Postgres
- dbt
- OpenAI
- Vercel AI SDK
Questions we get asked
Tell us the outcome, not the tooling.
Send us the situation you are in. We will tell you which discipline it belongs to, what we would do first and what it costs, including when the answer is to wait.