Startup
Background
A CTO of a 12-person AI startup building a developer tool. Ships weekly, hires monthly, relies heavily on AI agents for code generation, documentation, and planning. Constant context switching between features. Institutional knowledge lives in the heads of 3 founding engineers — and increasingly in forgotten Slack threads.
Goals
- Accelerate onboarding: new hires should find answers independently within their first week
- Capture architectural decisions before they’re forgotten in Slack
- Make AI agents codebase-aware so generated code matches project conventions
- Maintain visibility of priorities despite rapid pace
Pain Points
- Feature decisions made in Slack and forgotten within days
- Architecture knowledge siloed in founding engineers — bus factor of 3
- New hires can’t find answers: “Why Postgres?” “Where’s auth documented?” “What’s the current priority?”
- AI agents produce generic boilerplate that misses project conventions
How They Use aDNA
Heavy use of the knowledge and operations layers to capture institutional memory:
what/decisions/— 40+ architecture decision records (ADRs) capturing the “why” behind every technical choicewhat/context/— architecture overview, API conventions, tech stack rationale loaded by agents before code generation- STATE.md — updated weekly with sprint focus, blockers, shipping timeline
how/campaigns/andhow/missions/— quarterly goals decomposed into sprint-sized missionswhat/lattices/— CI/CD pipeline and deployment workflow captured as composable lattice definitions
The startup vault grows fast. Within 6 months it has more files than most mature vaults, driven by the pace of decision-making and feature shipping.
Self-reference: This vault’s what/decisions/ directory demonstrates the ADR pattern the startup relies on. Each decision file captures context, the decision itself, and consequences — exactly the institutional memory that would otherwise evaporate in Slack.
Typical Ontology Extensions
| Entity | Triad | Purpose |
|---|---|---|
api_spec | what/ | API endpoint documentation with versioning |
onboarding_checklist | how/ | New hire setup and orientation tasks |
customer_feedback | who/ | Structured customer feedback tracking |
Related
- Startup First Hour — 60-minute CTO quickstart: clone → CLAUDE.md → STATE.md → first ADR
- Startup Use Case — full narrative
- Tutorial: Build a Campaign — structuring quarterly goals
- Progressive Enrichment Pattern — growing the vault organically