Case study
Gridsafi
Mapping and screening for infrastructure investors in Africa. The product turns messy spatial inputs into something you can show in a meeting without apologizing for the tool.
The problem
Gridsafi sits where policy, finance, and geography meet. Buyers needed to see risk and opportunity on a map, but the inputs lived in too many places, each with its own gaps and refresh cadence.
The founders did not need a vision deck. They needed software that stayed calm when someone poked at the numbers in real time.
The challenge
Trust had a short half life. Every extra feature made stability worse. Every oversimplified map made credibility worse.
The stack also had to stay small enough that the team could run it without hiring a platform org.
What we did
We sat with the founders and split true screening flows from ad hoc research tangents, then built the smallest set of screens that still felt serious to infrastructure people.
We tightened how layers and tables joined, watched performance on real extracts, and made sharing explicit so calls did not end with “we will send a file.”
Where we polished, we biased toward plain language, sane defaults, and guardrails that reduce dumb mistakes more than flashy chrome.
Outcome
Pilot conversations had a real artifact behind them: repeatable steps, consistent maps, and fewer late nights fixing data before a demo.
The work stayed close to the actual job, screening and explanation, not generic “platform” talk.
Technologies used
- TypeScript
- Web mapping and geospatial workflows
- Data integration and validation
- REST APIs for partner flows
- Performance work on real world datasets
Key takeaways
- Most geospatial pain is in the joins: layers, tables, and what people say out loud in the room.
- Pilots need dull reliability more than novel UI.
- Small teams move faster when one person owns data through demo.