Insight Platform
ChatGPT can make a picture of a dashboard, but we build what makes it right every time.
This is for organisations that run on their data and already use AI with it. We build the foundations your own AI tools need to produce dashboards you can trust, along with a way for your experts to correct the results, so that it improves the more they use it.
Dashboard or platform
A chat window starts from nothing every time you ask it something.
If you ask an AI tool for a sales dashboard on Monday and again on Tuesday, you may well get two good-looking pictures with two different numbers. Neither will tell you where its figures came from, and you cannot click into either of them to see the detail.
Illustrative diagram
A dashboard compared with a platform
A dashboard from a chat window
A platform your AI builds on
What we build
We build five things, and the fifth is the one that matters most.
A clean data model, with provenance
We bring your data together in one model, with every figure traceable back to the system and the record it came from (its provenance), so that when someone asks "where did that number come from?", there is an answer.
Agreed metric definitions
What counts as revenue, an active customer or a late order is written down once and agreed by the people who own those figures. Every chart then uses the same definition, whoever (or whatever) builds it.
A reusable component library
We build the charts, tables and filters properly once, so that they are accessible, consistent, and able to drill down from a total to the rows behind it. Your AI then assembles dashboards from these rather than drawing new ones each time.
Instructions for your own AI tools
This is written guidance your AI assistants read before they answer, telling them which data to use, which definitions apply and which components to build with. It works with the tools you are already paying for.
A learning loop with your experts at the centre
Each time the AI produces a result, the people who know the subject say how confident they are in it, and why, if they are not. Those ratings are stored alongside the question, the data and the answer.
The system learns from them which approaches your experts trust, which definitions need tightening, and where the AI should ask rather than guess. It improves with use, and your experts stay in charge of what "right" means.
This is the part a chat window cannot offer: because the people who know your business are teaching it, the system becomes a little better every week.
Who it suits
It suits organisations that live by their numbers.
- Your analysts already ask AI tools for charts and summaries, and you are not sure the numbers agree.
- Different teams bring different figures for the same thing to the same meeting.
- You would like people to explore the data for themselves, without a queue of report requests.
- You have experts whose judgement matters, and you want AI to learn from them rather than work around them.
We have built this kind of system before, including AI pipelines where people review, rate and correct what the AI decided. You can read more about that in our work.
Production Readiness Review
In five days, we will tell you whether it is ready to go live.
The price is fixed before we start, from £1,500. You will receive a short report with a red, amber and green summary, and the ten fixes that matter most, in the order we would tackle them.