Agent-to-agent collaboration over A2A

Your LangGraph pipeline delegates to your CrewAI crew — over an open protocol.

error_outline

Without memory

Real organizations end up with agents on different frameworks — a LangGraph pipeline here, a CrewAI crew there, a custom bot in between. Getting them to hand work to each other means brittle bespoke glue, and none of them share what they learn.

check_circle

With Rekall

Rekall speaks the open A2A protocol: send a task to any Rekall-hosted agent, stream its progress over SSE, get push-notified on completion. And because every agent shares Rekall’s graph-backed memory, they collaborate through shared context — not just messages.

How it works

Every agent gets an Agent Card

Rekall publishes A2A agent cards with streaming and push-notification capabilities — any A2A client can discover and call your agents.

Delegate with message/send

JSON-RPC in, an A2A task out. Stream live status over SSE with message/stream, or register a webhook for push notifications.

Sleeping agents wake on message

Send a message to a dormant agent and wake-on-message spins up a run to handle it — no polling loop babysitting required.

Memory is the real collaboration layer

Delegated tasks carry context, and results land in shared memory — the delegating agent recalls what the delegate learned.

See it in action

Delegating a task over A2A

curl -X POST https://api.rekall.ai/api/v1/a2a \
  -H "Authorization: Bearer rk_..." \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0", "id": 1,
    "method": "message/send",
    "params": { "message": {
      "kind": "message", "role": "user",
      "messageId": "msg-001",
      "parts": [{ "kind": "text",
        "text": "Summarise this week's incidents" }]
    } }
  }'
# → an A2A task; stream it live with message/stream