01
Human context
The problem
Data work often spans collection, preparation, method choice, analysis, charts, dashboards, and reporting.
Specialist time and commercial tools can be expensive, and fragmented handoffs make reasoning harder to inspect.
02
What a person can do
User-visible capabilities
Data workbench
A person can import CSV or spreadsheet data, collect selected public data, inspect columns, and receive guidance toward supported analyses.
Reviewable outputs
The product can surface assumptions and effect estimates, build charts or bounded dashboard summaries, and export material for review or reproduction.
03
Broad context
Engineering context
- Analysis surfaces
- A local Python analysis engine and browser workbench provide constrained operations with on-device supported methods.
- Charts and exports
- Chart and dashboard surfaces pair with reproducibility exports for later review.
04
Human responsibility
Decision boundary
Local-first describes imported-data analysis, not every network action. The catalog is narrower than a full statistics package. Sampling, study design, data quality, interpretation, and expert judgment still matter. It is not universally better than SPSS or a replacement for a statistician or data scientist.
05
Claim boundary
Limits stated plainly
- Optional AI can help interpret a question while supported calculations remain in the analysis engine.
- The product does not replace expert judgment or a full statistics package.