How GenZAgents works

One line per developer. Every AI conversation your team has — across Claude, ChatGPT, Cursor, Cline, Gemini — becomes a signed, searchable, transferable receipt under your organisation's account. The trick is that we own none of the work; you do.

1. Sign in. Generate an API key. Register an agent.

Google, Microsoft, or GitHub OAuth. Your agent gets a cryptographic identity (a did:genz:...) backed by an Ed25519 keypair. That identity is portable across every LLM provider.

2. Drop the MCP server into every dev's tools.

# Claude Desktop / Cursor / Cline: one block in the MCP config
{
  "mcpServers": {
    "genzagents": {
      "command": "npx",
      "args": ["-y", "@genzagentsio/mcp-server"],
      "env": { "GENZAGENTS_API_KEY": "sk_live_..." }
    }
  }
}

For Claude.ai and ChatGPT web (where MCP isn't available), install our browser extension. The dev experience is identical: nothing changes about how they use the LLM. Receipts land on the dashboard silently.

3. Every conversation becomes a signed receipt.

Each receipt is JCS-canonicalised, Ed25519-signed, and tagged with:

  • Per-author attribution (which employee triggered the work)
  • Per-project (the conversation it's part of)
  • Per-provider (Claude, ChatGPT, Cursor, etc.)
  • Environment (prod, staging, dev — staging excluded from trust scores)
  • Runtime cost in £ + token usage
  • Optional session replay payload for high-stakes work

4. Continuity becomes automatic.

Four scenarios our MCP tools handle out of the box:

Cross-provider handoff

Alice spent 3 hours in Claude. Bob needs to continue in ChatGPT. He says "restore my Q3 forecast chat" — our restore_chat tool returns the portable manifest, Bob keeps working with full context. 90 seconds vs 30 minutes of pestering Alice.

Employee turnover

Sarah resigns. Her agent + all receipts + memory snapshots transfer to Tom via cryptographic countersign. Tom inherits the full work history, not a blank slate.

Chat → code

Dave's Claude Desktop spec continues in Claude Code seamlessly. Same MCP, same agent DID, same receipt pool. No re-explaining.

Org-wide de-dup

Maria opens Claude to draft SOC 2 audit prep. Our org_context_lookup tool surfaces: "Carlos finished this on May 3rd — start from his work?" One day of duplicate work avoided.

5. Compliance is the bonus, not the pitch.

Once you're using us for continuity, the SOC 2 / ISO 42001 / EU AI Act evidence packs auto-generate from the underlying receipts. We don't sell compliance — we sell the knowledge layer; compliance comes free because the audit trail is structurally inherent.

Try it in 5 minutes

Free tier covers 1k receipts/month. No credit card.