Agent Overview

The Agent Designer consists of multiple panels for designing and building your agents

Orchestrate your agent's flow by laying out behaviors and configuring actions

Interactively chat with your agent while you build it

Trace agent behavior, replay events against your latest changes & establish regression tests

Author prompts in a purpose-built text editor with built-in abstractions for conversation history, RAG and more

Optionally use Python to build your prompts, translate completions to actions, prepare API requests and handle API responses.

Access useful tools for building an assistant, such as ad hoc resource search

Configure your agent's Default LLM, Session Fields, Configs, panel order & more