Agent vs Human Inboxes
When to use each API surface and how they work together.
Agentry exposes two distinct API surfaces:
- Agent API (
/agent/v0) for AgentMail-compatible mailbox workflows. - Human API (
/human/v0) for connected human account workflows.
Decision Framework
Use Agent API when:
- You need drop-in AgentMail semantics and SDK compatibility.
- You are creating/managing hosted inboxes and agent-originated messages.
- You rely on agent events like
message.receivedandmessage.delivered.
Use Human API when:
- You need to connect real human-owned mailboxes (OAuth or IMAP).
- You need smart routing and relationship-aware context.
- You need deep memory/context operations tied to human account activity.
Namespace Summary
| Surface | Base URL | Primary use case |
|---|---|---|
| Agent | https://api.agentry.to/agent/v0 | Hosted inbox and AgentMail parity workflows |
| Human | https://api.agentry.to/human/v0 | Connected human accounts and context workflows |
Shared Building Blocks
Both surfaces share core product infrastructure:
- Authentication and permission boundaries
- Durable webhook delivery and retry behavior
- Realtime fanout model and delivery filtering
- Audit logging and usage accounting
Combined Pattern
Many products use both surfaces together:
- Receive and classify intent via Human smart routing.
- Pull relationship/context from Human endpoints.
- Generate/send operational responses via Agent hosted inboxes.
This keeps customer communication grounded in human context while preserving strict agent operation boundaries.