Upload your validated dataset CSV and get an ERD you can actually explore.
This app groups fields into tables, infers likely relationships, highlights fact versus dimension style tables, and lets you inspect columns, join candidates, and metadata without spelunking through a giant spreadsheet like an archaeologist with a caffeine habit.
TABLE_NAMECOLUMN_NAMEDATA_TYPEFIELD_TYPEBUSINESS_LOGIC_DERIVATIONJOB_CONTEXT_NOTESSOURCEUSE_CASESJOIN_NOTES
How the inference works
1
Explicit joins win
If JOIN_NOTES mentions another table, the app treats that as a high‑confidence relationship.
2
Semantic matches fill the gaps
Column names are normalised so keys like DEPT_ID can line up with DEPTID and similar dimension keys.
3
Confidence keeps the graph sane
Use the confidence filter to reduce noise when a dataset has lots of repeated job‑context patterns.
ERD graph
Click tables or edges to inspect details. Drag nodes around if you want to play architect.
Fact-like tables: 0Visible edges: 0