Work · AI tool · In use by implementation teams
AI-Assisted Data Mapping
Turns a customer's raw data file into a clean, template-ready file, with AI suggesting the column matches.
ReactTypeScriptGeminiSupabase
context
Built at work for implementation teams
The problem
Every integration starts with a customer file whose columns never match the template. Teams matched them by hand, row by row, and errors surfaced late.
What I built
- Upload a CSV, pick a mapping template, then map columns by hand or accept Gemini suggestions ranked by confidence.
- Validation rules check every row before export, so bad data is caught at the start instead of at go-live.
- An admin side for templates, taxonomies (domains, products, modules), field libraries, maturity levels, tags and validation rules.
- Built-in documentation and Docker deployment so other teams can run it.
How it works
Drop in the customer's CSV and choose the template it needs to match.
Impact
- Part of the toolset that cut typical tasks from about 5 hours to under an hour.
- Adopted by implementation teams for customer onboarding.
What I learned
AI suggestions work best with a confidence score and an easy override. People trust the tool because they can see why it chose a match.