A Series B UK fintech had already done the hard data work. Their analytics layer was in place. But reporting still depended on people manually building and maintaining dashboards in Sigma.
The bottleneck was not the warehouse. It was the production loop around the dashboard.
Bottl worked with the CTO to look at where the team was losing time. The principle was simple: keep humans close to judgment, and move the mechanical parts closer to code.
For BI, that meant reducing dependence on bespoke dashboard work and moving more reporting into lightweight Python and Streamlit apps. If the data already existed, the first usable dashboard could be shipped in minutes. Agents could help create the interface, manage deployment, summarize changes, and generate recurring PDF reports for executives.
The stack was intentionally plain: Slack, Python, Streamlit, scheduled jobs, service workers, and email reports.
The BI team still owned the reporting function. The difference was that senior analytics engineering time was no longer trapped in low-leverage maintenance and one-off requests. More of that time could go back into modeling, architecture, and decisions that actually moved the business.
The lesson was simple: AI is most useful when it removes the manual loop around good data. Not when it decorates the old dashboard process.