Durable intent execution for long-running operations.

• Webhooks fail silently or arrive twice
• Retry logic scattered across services
• Human approvals break automation chains
• No shared execution state across time
• Polling loops add load and hide failures

AXP is the Intent Protocol: submit an intent, observe a durable lifecycle, and receive a terminal outcome later.

• Not RPC. Create returns intent_id + state, not a synchronous result.
• Ordered lifecycle events. Per-intent monotonic sequencing enables replay and recovery.
• Explicit waiting states. Human/tool/agent/time are modeled as WAITING transitions.
• Handler transfer. Execution ownership can move without breaking lifecycle continuity.

Service A             AXME Cloud                    Service B
   |                      |                        |
   |-- submit intent ---->|                        |
   |                      |-- route via Registry ->|
   |                      |-- async execution ---->|
   |                      |<-- completion event ---|
   |<-- completion -------|                        |
   |                      |                        |
   +-- durable lifecycle across time (seconds to hours)

Independent services stay decoupled while AXME handles lifecycle, routing, and delivery semantics.

• Cross-boundary routing. Route intents across services and organizations via The Registry.
• Protocol, not framework. Integrate via SDKs or the versioned Public API (OpenAPI) without adopting a workflow runtime.
• Identity is first-class. Global addresses and deterministic route lookup before execution.
• Initiator can disconnect. Completion via stream, polling, or reply_to inbox — no callbacks.

pip install axme

$ python example.py
Intent submitted: ax_93ksd...
Waiting for completion...
✔ Report ready: https://axme.ai/results/ax_93ksd

No polling loops. No ad-hoc schedulers. No custom workflow glue code.

• At-least-once delivery with replay support for recovery
• Idempotency support for retry-safe writes
• Three continuation modes: live stream, polling fallback, and offline reply_to delivery
• Initiators can disconnect and resume without losing lifecycle progress

• Transport-agnostic protocol with deterministic route resolution
• Current managed transport uses Matrix with native E2EE
• Owner-scoped authorization with fail-closed JWT/JWKS validation
• Credential-safe logging and auditable access events

AXME supports AI agents as first-class execution actors within the same durable lifecycle model.

• Long-running agent workflows
• Human-in-the-loop AI operations
• Cross-agent coordination
• Deterministic execution for LLM-driven processes

Everything you need to integrate: examples, SDKs, docs, and protocol reference.

MCP is a supported integration path for assistant tools. AXP is the continuation and coordination layer — not interchangeable with MCP.

Auto-provisioned access. No sales calls. No contracts.

Access is granted instantly after submission.

Move to the onboarding screen to create your workspace and run the example.