ProductOS

Revolutionising product management.
What's shipped

Knowledge Autopilot

Ray and Lex answer product questions in Slack and Intercom, log uncovered ones, and a weekly digest drafts new help-centre articles. Runtime lives at knowledge/ in this repo · Deployed to Railway. Configure in Settings → Knowledge.

R
Ray AI
Product / QA expert
Answers "is this possible today? how?" using Intercom, Linear, Notion, GitHub, and Slack KB channels. Auto-watches configured channels.
Default model
claude-haiku-4-5-20251001
Escalates to
claude-sonnet-4-6
Invoke
@Ray AI or auto-watch
Manifest
knowledge/slack-manifest-ray.yml
L
Lex AI
Product strategist
Compares the request against this quarter's Linear roadmap and the strategic pillars in prompts/product-thinking.md. Asks sharpening questions when under-specified.
Model
claude-sonnet-4-6
Triggered by
Ray flagging NEEDS_LEX, or @Lex AI
Reads
prompts/product-thinking.md
Manifest
knowledge/slack-manifest-lex.yml
Knowledge channels

Ray searches recent messages in the channels listed under RAY_KB_CHANNELS (comma-separated channel IDs). He must be invited to each. Good candidates: #product, #support, #eng-help.

Lookback window controlled by RAY_KB_LOOKBACK_DAYS (default 90).

Cost guardrails
Hard ceiling per thread
MAX_COST_USD_PER_THREAD · default $1.00
Tool calls per agent
MAX_TOOL_CALLS_PER_AGENT · default 8
Average per thread
~20¢ (Ray clean: 3–8¢ · Ray+Lex prioritise: 30–55¢)
Ask Ray from Intercom

An Intercom inbox app surfaces an Ask Ray button on every conversation. Clicking it runs Ray against the conversation context and posts the answer back as a canvas reply — handy for support agents who want a second opinion before responding.

Setup
  1. In Intercom Developer Hub, create an inbox app.
  2. Set the Initialize and Submit URLs to https://<your-railway-domain>/intercom/canvas.
  3. Set INTERCOM_CLIENT_SECRET in Railway env so signature verification works.
  4. Adjust INTERCOM_CANVAS_PORT if the default 3002 conflicts.
  5. Install the inbox app on your workspace and pin it to the conversation sidebar.
Endpoint
Source
knowledge/src/intercom/canvas-kit.ts
Path
POST /intercom/canvas
Healthcheck
GET /intercom/health
Agent used
Ray (no Lex chain in canvas mode)
Gap finder
Compare the published help centre against product signals (releases, changelog, README, recent PRs) to surface articles you should write next. v1 — uses stubbed scan data; v2 will read data/kb-gaps.jsonl and the live changelog.