Ergon Documentation
Ergon is the operating layer for the AI-managed company. Self-governing agents and humans work side by side under the same identities, permissions, and audit — bounded when you want them bounded, autonomous when you want them autonomous.
Start with the Concepts to understand how this works. Use the Guides to learn how to build with each service. Dive into the API Reference when you need field-level detail.
Concepts
The six pieces of the worldview. Read these once and the rest of the documentation tells a coherent story.
The AI-managed company
The vision: an operating layer where self-governing AI systems plan and execute work alongside humans, under the same identity, permissions, and audit.
Identities & boundaries
Agents are first-class principals with their own grants. The two-way handshake and the autonomy spectrum keep agent work auditable without confused-deputy holes.
Agent memory
Episodic conversation notes and digests, transparent user preferences, and durable agent semantic memories.
Multi-agent collaboration
Conversations as the universal substrate. Human↔agent, multi-agent meetings, and agent-to-agent handoffs — emergent, not a separate protocol.
Event-driven autonomy
Agents react, they don't poll. Step graphs, triggers, depth limits, and the rule engine that turns business policy into autonomous operations.
Multi-agent choreography
Agents collaborate by sharing keys, not by calling each other. Five resume-key namespaces are the vocabulary of cross-agent, cross-workflow, cross-channel hand-offs.
Guides
Understand how each service works, how they connect, and how to integrate them.
Platform Overview
Understand the Ergon architecture, services, tenancy model, and how everything connects.
IAM
Identity and access for the whole platform — principals, companies, members, API keys, agent identities, and permission checks.
Agent Hub
How LLM agents work — configs, modes, conversations, invocation, and integration patterns.
Conversations
The dialogue substrate — conversation lifecycle, paginated messages, notes & summaries, privacy, and real-time streaming.
Automations
React to events with step graphs — triggers, runs, resume keys, and cross-service connections.
Events & Trigger Catalog
Browse every automation event, payload schema, resource scope, and compatible predicate operator.
Workflows
Phases, routing, items, automations, templates, and public forms.
Channels
Multi-channel messaging, email setup, inbound triggers, and agent routing.
Buckets
Document ingestion pipeline, AI processing, semantic search, and cross-service sharing.
Worksheets
Spreadsheet-style data with typed columns, views, formulas, and automations.
Event Streams
Centralized event store — query event history, metrics, distributed traces, and stream updates in real-time.
Compute
HTTP Collections for prepared API calls, governed egress, and Vault-backed credential indirection. Functions and Browser are coming soon.
Vault
Namespaces and credentials for secrets. Server-side resolution and crypto, with secrets that never leave the boundary.
Permissions
RBAC model, principals, roles, resource patterns, and cross-service grants.
Webhooks & Events
Real-time events, WebSocket streams, channel triggers, and payload formats.
API Reference
Detailed endpoint documentation with request/response schemas and code examples.
IAM
Central identity and access management — companies, members, roles, permissions, API keys, agents, and service access control for the entire platform.
Agent Hub
Configure and run AI agents with LLM provider keys, model configs, tool access, and real-time agent invocation with streaming responses.
Conversations
The universal substrate for dialogue — human↔human, human↔agent, and agent↔agent. Identity, permissions, audit, and real-time streaming built in.
Automations
The reaction substrate. Subscribe to events, run step graphs that invoke agents, send messages, transition workflow items — autonomy expressed as policy.
Workflows
Kanban-style workflows with phases, fields, routing, items, assignment groups, templates, automations, and public forms for external submissions.
Channels
Multi-channel messaging with email configs, addresses, outbound send, inbound triggers, address-level ACL, and cross-service grants.
Buckets
Document management with folders, file uploads, AI-powered text extraction, chunking, and vector-based semantic search for RAG.
Worksheets
Spreadsheet-style data with typed columns, rows, saved views, CSV/XLSX import and export, automations, and real-time collaboration.
Event Streams
Centralized event store that ingests events from all services. Query event history, metrics, distributed traces, and stream updates in real-time.
Compute
Run prepared HTTP Collections and one-off API calls with layered egress control and Vault-backed credential indirection. Functions and Browser families coming soon.
Vault
Store, encrypt, and resolve secrets through namespaces and credentials. Server-side resolution and cryptographic operations keep secrets out of every response.
Authentication
All API requests require a Bearer token. Authenticate using API keys with the client credentials flow.
Learn about authentication →Base URL
All services are accessible through a unified gateway at https://platform.ergondata.ai. Each service has its own path prefix.