Product guide

Using the app

A practical guide to moving from a raw file to evidence your team can inspect, export, and share.

Analyze a file

Drop up to 10 files in the Ridge workspace, or paste an https:// link to a file. Supported: .xlsx .xls .csv .json .txt .md .pdf .pptx .ppt .docx .doc.
Uploads are capped at 4 MB per request; larger files (to 25 MB) go through the link box, which fetches server-side.
Multi-sheet workbooks get a sheet picker after the first run.

"Analyze data" runs the deterministic pipeline: statistics per column, a data-quality grade, pairwise-complete correlations with sample sizes, date profiles, and structured evidence.
Adding an Anthropic API key through AI settings unlocks Explain with Claude, cross-file synthesis, and follow-up questions. Deterministic numbers stay exactly the same either way.
For documents (PDF, Word, PowerPoint), keyless mode shows the extracted text and counts; AI reading requires a key.

Every material finding is a structured object: claim, metric, columns, method, sample size n, coverage, strength, and caveat. AI text may explain these objects but cannot add numbers of its own.
Optionally pick a target column on the dashboard to focus evidence on one outcome: group comparisons, correlations with the target, missingness impact, and trends over time.

Nothing is stored unless you tick Save this analysis before running — see Privacy for exactly what a saved analysis contains.
Saved analyses appear under Recent analyses, produce a share link, and can be deleted there.
Any result can be exported as JSON or as a printable report, saved or not.

Architecture, API reference and self-hosting instructions live in the repository: github.com/MohammedAlkindi/Ridge.