Evidence-Based Planning
Traditional AI documentation tools often suffer from "hallucinations" because they lack context. Docxa solves this via Evidence-Based Planning.
What is Evidence?
Evidence is simply structured context that the AI needs to write a specific section of a document. Docxa recognizes several types of evidence:
- Business Context: Goals, target audience, problem statements. (Usually from interviews).
- Technical Context: Frameworks, architecture patterns, dependencies. (Usually from repo analysis).
- Existing Docs: Upstream documents (e.g., a BRD provides evidence for a PRD).
The Readiness Check
When you run docxa generate --plan, Doxa evaluates every section of the target template against its Evidence Requirements.
| Requirement | Satisfied By | Status |
|---|---|---|
business_context |
PRD Interview | ✅ |
arch_patterns |
Repository Analysis | ✅ |
ux_flow |
Missing | ⚠️ |
Generation Modes
- Strict: Blocks generation if any required evidence is missing.
- Flexible: Warns you about missing evidence but proceeds anyway.
- Assisted: Identifies missing context and suggests specifically which interview role to run to fill the gap.