Core concepts of the Inkeep Agent Platform

Copy page

Learn about the key building blocks of Inkeep - agents, graphs, tools, data components, and more.

Key concepts

Here's a high level of the key concepts of building Agents with Inkeep.

  • Agents: LLM and AI driven decision-making units defined by instructions ('prompts') that can take actions, answer questions, and coordinate with other agents as needed.
  • Graphs: A team of agents that work together to perform complex tasks across your apps and APIs.
  • Tools: The capabilities that agents can invoke. Tools can be your own custom functions or you can leverage our pre-built MCP tools for common integrations.
  • Transfers and Delegations: Relationships between agents. Transfers involve a transfer of control of the chat from one agent to another, and delegation allows an agent to create a sub-task for a different agent.
  • Traces: A log that gives you full visibility into every step of how an agent graph ran. Accessible via UI or OpenTelemetry.

Advanced features

  • Data Components: Define what agents can output in their messages so they can render rich, interactive UIs (lists, buttons, forms, etc.).
  • Context Fetchers: A way to retrieve and cache data about your user or apps so that an agent has it immediately available in its prompt.
  • Artifacts: A piece of data that can be saved by an agent so it is accessible to other agents or users.