Core Concepts

Chainabit's domain model is built around two layers: an AI execution layer for running agents, sessions, and tool calls, and a productivity layer for organizing tasks. Every concept connects back to the account and workspace that owns it.

The Domain Model

Account & Workspace

An Account is an individual developer or enterprise organization. Each account has one or more Workspaces — isolated environments for organizing work. Individual accounts get a default workspace; enterprises can create multiple with separate members and permissions.

Chainer

A Chainer is the person (or service account) using the platform. Every authenticated user is a Chainer. A Chainer owns agents, runs sessions, and manages workspaces.

AI Agent

An AI Agent is a configurable autonomous entity that can hold conversations, execute tools, and query knowledge. Agents come in two forms:

• Twin — a persistent AI identity with long-term memory. A Twin accumulates context across sessions and evolves with use. Think of it as an AI coworker.
• Definition — a reusable agent configuration (model, system prompt, tools, skills) that can be instantiated on demand for specific tasks.

See Agents API and Twins API.

Session & Messages

A Session is a persistent conversation thread between a user (or automation) and an AI agent. Sessions hold message history, tool call results, and execution state.

Messages are the unit of exchange within a session:

• User message — input sent to the agent
• Assistant message — the agent's reply, which may include tool call results
• System message — injected context (usually from the agent definition)

Responses are streamed via server-sent events (SSE). See Streaming and Sessions API.

Context (Knowledge)

A Context is a scoped knowledge store that an agent can query for retrieval-augmented generation (RAG). Contexts hold documents, embeddings, and structured data indexed for semantic search.

Agents query contexts automatically when answering questions in their domain, or explicitly via the contexts:read API scope. See Contexts API.

Connectors

Connectors link agents to external services — Slack, Notion, GitHub, Stripe, Airtable, Linear, and more. Each connector exposes a set of Tools that agents can call during a session (e.g., "create a Notion page", "list open GitHub issues").

Connectors are authenticated via OAuth or credential storage and scoped to a workspace. See Connectors API.

Chainy & Bits (Productivity)

The productivity layer organizes task work within a workspace:

• Chainy — a goal-scoped micro-workspace. Contains direct tasks (Bits) and a kanban board for a specific project or objective. Think of it as a lightweight project container.
• Bit — an atomic task event. A Bit starts as a pending task (with title, priority, and stage) and transitions to a completed record when done.

See Chainies API and Bits API.

API Keys & Developer Access

Service-to-service integrations authenticate using API keys (chb_sk_*). Keys are scoped to the minimum permissions required:

Scope	Access
contexts:read / contexts:write	Knowledge context operations
agents:read / agents:execute	Agent reads and session execution
analytics:read	Analytics and audit data
members:read	Workspace member reads

API keys require the Chainer subscription tier or above. See Developer Access.