> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.moveworks.com/agent-studio/guides/architecture/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.moveworks.com/_mcp/server.

# Architecture

## Docs

- [Building a Conversational Plugin](https://docs.moveworks.com/agent-studio/guides/architecture/building-a-conversational-plugin.md): End-to-end reference architecture for a multi-step conversational plugin.
- [Building an Ambient Agent](https://docs.moveworks.com/agent-studio/guides/architecture/building-an-ambient-agent.md): End-to-end reference architecture for a webhook-triggered ambient agent with cross-system orchestration.
- [Choose how to connect external systems to Moveworks](https://docs.moveworks.com/agent-studio/guides/architecture/choose-how-to-connect-agent-studio.md): Compare the ways to connect external systems to Moveworks using enterprise search, MCP servers, plugins, or ServiceNow.
- [LLM Orchestration Patterns](https://docs.moveworks.com/agent-studio/guides/architecture/orchestration-patterns.md): Reusable patterns for composing Compound Action primitives and LLM Actions into reliable LLM workflows.
- [Fan Out and Synthesize](https://docs.moveworks.com/agent-studio/guides/architecture/orchestration-patterns/fan-out-and-synthesize.md): Query several systems in parallel, then use an LLM Action to turn their combined results into one answer, like a summary or a recommendation.
- [Categorize and Route](https://docs.moveworks.com/agent-studio/guides/architecture/orchestration-patterns/categorize-and-route.md): First classify the broad domain, then run a second classification limited to domain suboptions.