DISTILL AGENT
Raw data goes in. Pure data comes out.
Most projects drown in messy, undocumented decisions. Distill filters the noise — walk through structured decision cards and export crystal-clear data with a full audit trail. No code, no residue.
Data files
Supported: CSV, TSV, Excel, Parquet, JSON, SPSS (.sav), Stata (.dta), SAS, ZIP (extracted automatically). Images and other non-tabular files are kept as companion files.
Drop data file(s) here
or click to choose — you can add more than one
Documentation optional
Codebook, data dictionary, or README — the agent reads it to understand your variables and load the data correctly.
Drop documentation here
PDF, Markdown, or plain text
Notes optional
Study description, known issues, outcome variable, etc.
or
What you get
✦
Clean dataset
CSV, Parquet, or Excel — ready for modelling or sharing.
⟳
Reproducible code
Standalone Python script that replays every decision.
▤
Audit report
Readable PDF + Markdown with every change justified.
◉
Decision logs
Structured JSON — feed to another AI or share with collaborators.
Inside the session
◌
Chat assistant
Ask questions about your data or decisions mid-session.
⊞
Variable inspector
Live distribution, missingness, and per-column stats.
◎
Zero coding
All cleaning logic handled — no Python required.
⚖
Human-in-the-loop
Every judgment call surfaces as a decision card.