Decision intelligence · built for real-world files
Ridge turns operational data into computed findings, clear quality signals, and an audit trail your team can follow. AI stays optional—and never invents the numbers.
no signup · no API key · source files never stored
global_revenue.csv · source
| month | market | revenue |
|---|---|---|
| 2026-01 | EMEA | 48,200 |
| 2026-02 | EMEA | 51,900 |
| 2026-03 | EMEA | 67,400 |
| 2026-02 | APAC | ∅ |
| 2026-03 | APAC | 30,800 |
Ridge · computed output
revenue ↔ spend rise together +0.91 · n=90 · 98.9% coverage
EMEA revenue grew 30% into March while APAC held flat—though the missing February figure caps confidence in the regional trend.
01
Spreadsheets, CSV, JSON, PDF, Word, PowerPoint — or paste a public link. No account, no sign-up, and no API key to get started.
02
Statistics, correlations with their sample sizes, a data-quality grade and structured evidence — all calculated outright, with no model involved.
03
Optional. Paste an Anthropic key to have the findings written up and to ask follow-up questions — always grounded in the numbers already computed.
Missingness, type consistency, IQR outliers, duplicate rows, and a weighted health grade are calculated in code — the AI reads them, it doesn't invent them.
A distribution, frequency or trend chart for every column, a scatter plot behind every correlation, and a correlation matrix — aggregated across the complete file, never inferred from preview rows or generated by a model.
No account. Files are parsed in memory and never written to disk. Nothing is saved unless you tick the box that saves it. The code is open — read it.
Open any finding to see its formula, inclusion rule, excluded-row count, and a bounded excerpt of the exact source rows involved.
Compare two files for schema drift, quality movement, mean intervals, and full-distribution shifts—without asking a model to do arithmetic.
Mean intervals, Welch tests, Cramér's V, and KS shifts carry sample sizes, p-values, effect sizes, and multiple-testing caveats.
For the teams expected to make a decision from a spreadsheet they did not build.
You get an export from a system you don't control. Ridge tells you what is actually in it — how complete each column is, what correlates, what looks wrong — before you build anything on top of it.
You need a defensible read on a file today, not a notebook and an afternoon. Drop it in, get the numbers, and know which ones rest on enough data to quote.
Profile a handoff before it reaches a warehouse or dashboard. Surface mixed types, missingness, duplicates, outliers, and weak statistical support early.
Ridge is open source and ships as a non-root, health-checked container. Keep the deterministic path fully internal, add your own Anthropic key when policy allows, and put the documented API behind your existing gateway.
Read the deployment guide →Uploads are parsed in memory and discarded when the request ends. Nothing is written to disk, and there is no database row unless you ask for one.
History and share links are off until you tick Save this analysis. Untouched, a run leaves nothing behind — and saved runs can be deleted.
An Anthropic key is optional and kept for the session by default. It travels with your request to Anthropic and is never stored server-side or logged.
Bring a real spreadsheet, or inspect a complete analysis in ten seconds.