Agents#

Roboto lets you run purpose-built agents on your robotics data. An agent can search your data, explore datasets and files, analyze topic data, summarize and triage datasets, and create events — all through natural-language conversation, grounded in the tools and permissions of your Roboto account.

Agents complement actions: an action runs your own containerized code deterministically, while an agent applies AI judgment using Roboto’s built-in tools. Reach for an action when the job is repeatable data processing. Reach for an agent when the work needs analysis, search, or judgment.

Four core components make up the agent surface:

  • Agents and Threads — an agent is a purpose-built AI worker you define once and launch repeatedly; a thread is a persistent, multi-turn conversation with the AI.

  • Goals — structured outcomes (like “triage this dataset”) that the AI must achieve before a turn completes.

  • Skills — your team’s stored procedures, which the AI can apply automatically or on demand.

  • MCP Connections — tools from external services, brought into your conversations via the Model Context Protocol.

MCP also works in the other direction: the Roboto MCP Server brings Roboto’s tools to external AI apps like Claude Code, Claude Desktop, Codex, Cursor, and VS Code.

Where you can use agents#

  • Web app: open AI Chat (the robot icon on any page) for ad hoc conversations, or the Agents section for threads, agents, and skills.

  • CLI: start an interactive session with roboto chat start.

  • Python SDK: drive threads programmatically with AgentThread.

On this page