Agent product

Nori

Job search infrastructure built around a career graph.

How it works
  1. Your history
  2. Career graph
  3. Match
  4. Tailor
  5. Apply

The problem

Most auto-apply tools hold a flat resume and a prompt. They rewrite your history to fit each job, and nobody can tell which claims are real.

How it works

Nori turns your history into a graph: roles, projects, skills, metrics and education as nodes, with edges like used in, achieved at and worked at. It finds roles across Greenhouse, Lever, Ashby and SmartRecruiters boards, researches each company with citations, tailors a one-page resume where every bullet traces back to a fact about you, fills ATS forms with review-before-submit, reads verification codes from your mail and tracks replies.

“What am I missing for this role?” becomes a graph diff. You can drive it from Slack, Telegram, WhatsApp, email, a dashboard or MCP.

Built at the Cline × Composio × Parallel hackathon during Open Source AI Week in San Francisco, October 2026.

03 /

Key decisions

  • Tailoring is retrieval, not generation

    The job description is matched against your subgraph, and bullets are rewritten only from evidence nodes. Nothing is ever invented.

  • Review before submit

    Nori fills ATS forms but waits for approval, and the diff you approve shows which node each change came from.

  • The model is part of the product

    The operator sets one model endpoint for everyone, so there are no keys to paste. Self-hosting can run on a local Ollama, or on rules alone.

04 /

Results

ATS sources
35K+
Fact-backed tailoring
Graph
Optional local mode
Local