Alive developer documentation
Alive helps you find the right evaluation for a healthcare innovation at whatever stage it is in. You describe what you have; Alive works out what can be evaluated, runs the evaluations that apply, and tells you exactly how much each result can be trusted.
Status: Alive is in active development. The V0 app and API build locally; a hosted internal pilot has not been verified yet. Pages marked implemented describe code in this repository, not a live hosted service. Pages marked beta may change; planned features are not documented as if they work.
Choose your path
| You have… | Start with |
|---|---|
| An idea, a paragraph, or a document | Tutorial: from idea to evaluation |
| A workflow you want to simulate | Tutorial: run a synthetic world |
| A question about a change you are considering | How-to: author a world from a question (beta) |
| A private repository or complex project | How-to: use the AI intake prompt |
| Code that should call Alive | HTTP API reference and CLI reference |
| An IDE or terminal coding agent | Installed-client agent guide |
Client library references: Python and TypeScript. Packages are buildable artifacts and have not been published to PyPI/npm yet.
Core ideas in one minute
- IntakeSpec describes what your project is and what each statement rests on. Every project has one. Reference
- EvaluationPlan is Alive's decision about which evaluations apply, which do not, and why. Reference
- WorldSpec is an optional, executable synthetic world. Many projects never need one. Reference
- BusinessAssessment names commercial claims, hypotheses and unknowns without a viability score. Reference
- Results are always typed: verified failures observed in a trace, modeled outcomes that depend on stated assumptions, and AI hypotheses to test. Explanation
Alive's worlds are simulation models, not digital twins, and nothing Alive produces is evidence of clinical benefit. Read simulation validity before relying on a number.