Analyze
Analyze is our AI-assisted research workspace for biological and medical data. Upload a dataset, describe the analysis you want in plain English, and an AI agent writes, runs, and documents the statistical code for you. Results are versioned and reproducible — anyone you share the inquiry with can see exactly how every figure and table was produced.
Access is by invitation during the beta. Request an account.
How it works
1 Upload your data
Upload one or more Excel or CSV files to your private Data Library. Files are stored encrypted and only visible to you unless you choose to share them.
2 Prepare a derived dataset
Tell the data-prep agent how you want your raw files cleaned, joined, and transformed. It writes a Python script, runs it in an isolated sandbox, and shows you a preview of the result. Iterate until the dataset looks right, then publish it to your library.
3 Run analyses
Open an Inquiry against your prepared dataset and create as many analyses as you need — a Table 1, a regression, a survival model, a figure, anything. Describe each one in chat and the analysis agent generates and runs the code, returning tables and plots.
4 Share & reproduce
Publish an inquiry to lock down a reproducible record: dataset version + script + results + auto-generated methods description. Share with collaborators by email, or download a self-contained reproducibility bundle.
What makes it different
- You own your data. Datasets live in your private library. Sharing is explicit and per-inquiry.
- Sandboxed execution. Agent-generated scripts run in isolated Docker containers — they cannot reach your other datasets or the underlying server.
- Reproducible by construction. Every result is paired with the exact script, input version, and parameters used to produce it. Re-run any analysis with one click months later and get the same answer.
- Methods text included. The agent drafts a Methods-section paragraph describing what was done, ready to drop into a manuscript.