Engineering

Stop losing decisions in Slack threads.

The average engineer spends 30% of their time searching for information. Architecture decisions, deployment steps, and tribal knowledge live in people's heads.

The problem

Someone asks "why did we choose Postgres over Mongo?" and the answer is buried in a Slack thread from 2023. Deployment runbooks are outdated wikis. New engineers take months to understand the system because context is scattered everywhere.

How it works

Here's how you build your knowledge graph step by step using the ai CLI.

Map your codebase

Automatically turn your source code into searchable knowledge nodes. AST-based chunking captures functions, classes, and modules.

ai map ./src

Parses your code at the function level — search "retry with exponential backoff" and find the implementation.

Add an architecture decision

Capture the why behind technical choices. These are the decisions that are hardest to reconstruct later.

ai add "Chose PostgreSQL over MongoDB for the orders service \
  because we need strong transactional guarantees for \
  payment processing. Document store flexibility was \
  less important than ACID compliance for financial data." \
  --title "Why PostgreSQL for Orders" -t decision -d technology

Add your deployment process

Document the steps so they are searchable and consistent, not locked in one person's terminal history.

ai add "1. Create PR against main, require 1 approval. \
  2. CI runs: lint, test, build, security scan. \
  3. Merge triggers staging deploy via Cloud Run. \
  4. Smoke tests run automatically. \
  5. Manual promotion to production via 'deploy-prod' label. \
  6. Canary for 15 min, then full rollout." \
  --title "Deployment Process" -t process -d technology

Add an incident response runbook

The knowledge your team needs at 3am shouldn't be in someone's head.

ai add "Database connection pool exhaustion: \
  1. Check active connections: SELECT count(*) FROM pg_stat_activity \
  2. Identify long-running queries: check pg_stat_activity.state \
  3. If > 80% pool used, restart API pods (kubectl rollout restart) \
  4. Root cause: usually missing connection.release() in error paths" \
  --title "Runbook: DB Pool Exhaustion" -t process -d technology

Add API design conventions

Capture the patterns your team follows so code reviews are faster and AI agents generate consistent code.

ai add "REST endpoints: plural nouns, kebab-case. \
  Always return { data, meta } envelope. \
  Pagination via cursor, not offset. \
  Errors: { error: { code, message, details } }. \
  Auth: Bearer token in Authorization header. \
  Versioning: URL path /v1/, not headers." \
  --title "API Design Conventions" -t policy -d technology

Link decisions to code

Connect architecture decisions to the modules they affect so context travels with the code.

ai link abc123 def456 --rel "implemented-by"
ai link ghi789 jkl012 --rel "governs"

"Why PostgreSQL" is linked to the orders module. "API Conventions" governs all API route modules.

Query with full engineering context

When debugging or building new features, AI assembles all relevant context from your graph.

ai context "How does the orders service handle failures?"

Returns the architecture decision, deployment process, runbook, and relevant code — all connected.

Your engineering knowledge graph is live

Architecture decisions, deployment processes, and tribal knowledge are now searchable and connected. AI coding agents have real context about your system.

New engineers ramp up in weeks, not months

AI coding agents generate code that follows your conventions

Architecture decisions are discoverable, not buried in Slack

Runbooks and processes stay current and connected to code

Start building your knowledge graph

Free during beta. No credit card required.