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

# Routing

> Build once. Connect to every API, tool, and LLM. Ship products and agents to production without owning the integration stack.

## Docs

- [Models](https://docs.merge.dev/merge-workforce/models/overview.md): The four decisions behind employee model access, what an employee's AI client needs to route through Gateway, and what a request passes through on the way to a provider.
- [Routing](https://docs.merge.dev/merge-workforce/models/routing.md): What a routing policy is, the strategies available in the Workforce dashboard, how a policy attaches to a Group or becomes your organization default, and the model alias each policy mints.
- [Model access](https://docs.merge.dev/merge-workforce/models/access.md): How the All models, Blocklist, and Provider keys tabs relate, how to block a provider or model organization-wide, how to narrow access for one Group with its approved-models list, and what a refused request looks like.
- [Model guardrails](https://docs.merge.dev/merge-workforce/models/guardrails.md): DLP rules on prompt and completion text, prompt injection screening on the direct and indirect axes, the actions each can take, and where detections land.
- [Advanced](https://docs.merge.dev/merge-workforce/models/advanced.md): The controls on the Advanced tab that apply to every request at once: region and vendor restrictions for data sovereignty, plus zero data retention, request tags, model migrations, and context compression.
- [Budget](https://docs.merge.dev/merge-workforce/models/budget.md): How Workforce budgets work: a hard cap per Group written to its Gateway project, the organization number you plan against, and the levers that reduce spend without capping it.