Knowledge Engineering for Runtime Norms

Your AI agents don't know your org.

KERN encodes your senior engineers' operating rules into a schema your agents read on every PR. Two weeks. Fixed fee. Receipts in YAML.

Book a call

kern-skills.yaml v0.3

- id: sk-001
  name: No raw SQL in service layer
  trigger: agent generates SQL outside /repository/
  enforcement: reject + suggest repository pattern
  rationale: MAS audit trail at single boundary
  bad:  db.Exec("INSERT INTO transactions...")
  good: repository.Transactions.Insert(...)

01 — The encoding crisis

For thirty years, a senior engineer's correction on a PR taught the junior. Six months later, the junior caught it on someone else's PR. Knowledge moved through hands typing comments.

Now AI agents do the work. They read Confluence — out of date. They do not read your senior's mind. They write SQL in your service layer. Your senior corrects them. The next agent on the next PR makes the same mistake.

The agent never absorbs anything. It can't.

Tacit knowledge worked when humans did the work. Agents need it explicit, machine-readable, present at runtime.

02 — How it works

Encode

Four hours with your senior engineer. We capture 15–25 of the rules they correct most. Output: kern-skills.yaml in your repo.

Run

The schema lives in your repo. Your agents — Cursor, Claude, Devin — read it on every PR. We watch the firing pattern, tune the rules.

Compound

Every other KERN install surfaces patterns that flow back to your schema. Customer 5 starts at v0.4, not v0.0. Patterns cross. Data does not.

03 — The Wedge Sprint

Your seniors' rules. Your repo. Your agents.

BeforeA pre-sprint baseline lands on your sponsor's desk: DORA snapshot, AI ROI baseline, encoding-gap audit. One PDF, before the engagement begins.
DuringThe schema fires on real PRs in your repo. Senior-correction frequency starts trending. Live mid-sprint review with your squad lead.
AfterTelemetry report, encoding-gap map across teams, schema versioned in your repo, and a one-page board update on AI ROI you can defend.

Fixed fee · time-bounded · full credit toward your license when you continue

Book a call

04 — Founders

Naren Katakam

Architect

Eighteen months operating as a single-practitioner AI-native delivery organization. Designed the schema. Dogfoods it daily.

Lim Wei Ming

Delivery

Fifteen-plus years industrializing AI platforms in MAS-supervised banking. Hardens KERN for regulated production.

Teck Kiang Seet

Commercial

Twenty years in APAC enterprise services. Network spans CIOs and managing partners across Singapore and the European consulting axis.

05 — First move

A few Wedge Sprint slots open this quarter.

A short call to walk through what your seniors correct most. No deck. No follow-on commitment.

Book a call