Does Knit have an MCP Server for Outlook out of the box?
Yes — the Outlook MCP Server is live and already built, not a roadmap item. It exposes Outlook's real actions as AI-ready tools your agent can call directly.
Does the Outlook MCP Server support both reading and writing data?
Yes. Its actions cover both reads and writes, so your agent can pull data from Outlook and push updates back to it, not just query it.
How does authentication work for the Outlook MCP Server?
Knit manages Outlook's OAuth, credential storage, and refresh for you — there's no auth to configure yourself before your agent can start calling tools.
What happens if Outlook changes their API?
Knit absorbs it. We maintain and update the Outlook connector on our end, so the MCP Server keeps working without you tracking Outlook's changelog.
Do you store Outlook data or customer PII?
No. Knit acts as a stateless passthrough — data is processed on our application server and returned directly to your agent, without being stored on Knit's servers.
What is the pricing model, and is there a free trial?
You choose the model that fits your business — pricing based on integrated accounts, or based on API call volume. Every plan includes a 30-day free trial before billing starts.
Can I use this as an LLM tool for function calling?
Yes — this MCP server exposes Outlook's actions as tools your AI agent can call directly, the same underlying capability previously offered as a standalone LLM Tool integration. MCP is the protocol; the tool-calling access is the same.
Does this work with Claude, GPT, and other LLMs?
Yes — any MCP-compatible client or agent framework can connect to this server and call its exposed actions, regardless of which underlying LLM is driving the agent.