Company Kernel

Your AI is guessing about your company.

Map how your company really works — who does what, who approves what — in one place, and connect the AI tools you already use. Every answer is cited, or an honest “I don't know.”

Never a guess.

Illustration of Claude, Microsoft Copilot, ChatGPT, and Cursor connecting through MCP to a Company Kernel map and a cited sample invoice-approval answer.
Illustrative product map. One company map gives each connected AI only the rules it may see — and sources for the answer. Open full-size product map

Production workspaces require the one-time code your operator sent you.

A one-time invite your operator bound to this email may send a receipt; always save the on-screen copy before closing this tab.

lowercase letters, numbers and hyphensstarts with a letter2–63 characters

company-kernel-saas.fly.dev/yourco/studio/yourco/mcp

Works with Claude Cowork, ChatGPT, Microsoft Copilot and Cursor — or any other MCP client.

Free to start. No credit card. Production access uses a one-time invite.

Every answer comes with receipts.

Ask a question through any connected tool. The answer either traces back to a rule you wrote — or you don't get an answer at all.

When the rule isn't there, it says so. When two rules disagree, it shows both. When a key shouldn't see something, it doesn't.

public product proof · Atlas example company 
▸ company_ask "Who can issue an invoice?"

{
  "status": "answered",
  "answer": "The Controller. Issue invoice is
    step 1 of Invoice to cash and requires
    accounting.invoice.write and email.invoice.send.",
  "citations": [
    "controller —performs→ issue-invoice",
    "issue-invoice —requires→ invoice.write",
    "invoice/issue —guarded-by→ delivery-acceptance"
  ]
}

▸ company_ask "What discount can we offer?"

{
  "status": "unknown",
  "reason": "No policy covers discounting.",
  "hint": "Model it once in Studio — every
    connected tool learns it."
}

▸ company_ask "Who approves overtime?"

{
  "status": "conflicted",
  "sources": [
    "policy/field-ops → the Dispatcher",
    "policy/finance  → the Controller"
  ],
  "hint": "Two rules disagree. It shows both —
    it doesn't pick one for you."
}

▸ company_ask "What's our payroll approval chain?"

{
  "status": "not_authorized",
  "reason": "This key is scoped to Dispatcher.
    Payroll isn't in its view."
}
  • answered— cited to your rules
  • unknown— says what's missing
  • conflicted— shows both sources
  • not_authorized— that key can't see it

Four ways it can answer. Guessing isn't one of them.

Workspace live in two minutes.

  1. Create your workspace

    Name your company, choose an address, and it's live: Studio — where you map your company, in the browser — and your own endpoint at /yourco/mcp. Your private workspace starts with a safe governance scaffold and a guided first change; the Atlas company above is a public product proof, never sample facts mixed into your company.

  2. Connect an AI tool

    In ChatGPT, Copilot, or Cursor: paste your workspace URL and key into the tool's connector settings — no terminal needed. Using Claude Code? It's one line:

    claude mcp add yourco \
  https://company-kernel-saas.fly.dev/yourco/mcp \
  --transport http \
  --header "Authorization: Bearer ⟨your MCP key⟩"

  3. Ask something your company should know

    “Who can approve a purchase?” A cited answer means it's working. An unknown tells you exactly what to map next.