Upload a spreadsheet and dsbuddy profiles it, models it, and tells you what actually matters — in plain English, in under a minute.
200+ statistics · 3 models on a held-out test set · one verdict you can act on
Feature relationships
Rows analyzed
7,043
Best model
86%
Model comparison
Top finding
contract_end_date is 94% correlated with the outcome — likely leakage.
Drag a CSV, XLSX, or Parquet in. That's the only manual step in the whole pipeline.
customer_churn.csv
Before anything runs, we check structure — types, headers, encoding, duplicate columns.
Claude reads every column and tags what it actually means — currency, category, identifier, date.
Reading tenure_months, monthly_charges, contract_type…
Your data streams through profiling, correlation, and modeling stages — live, not a black box.
Three models train in parallel, each scored on a 20% test set they never saw during training.
Claude reasons over the full profile and ranks what actually matters — and what to fix first.
Everything lands in one dashboard — charts, risk flags, and a place to ask follow-up questions.
Built for people who’d rather understand their data than wrangle it.