Steady runs in the background by design. You don’t have to live in our app to get value from it. Agents do the work of distilling updates and activity, delivering personalized context to the tools you’re already living in like Slack, MS Teams, and email. Our MCP server, now rolling out in beta, is the next chapter in the same story: Steady inside the AI assistants and MCP-aware tools that you already use.
Why MCP?
For many customers, our apps for Slack and Microsoft Teams are the primary way they interact with Steady. But they’re limited by nature. Each app is a dedicated engineering effort that does exactly what we sit down to build, and nothing else. Want your digest summarized in a specific way, combined with data from another app, or piped into a workflow we didn’t anticipate? You’d have to wait for us to ship a feature for it.
Those limits get sharper as AI assistants take over more of people’s workdays. Our customers are spending more time in tools like Claude & ChatGPT, and working with an ever-expanding constellation of custom agents powered by tools like OpenClaw. These tools are general-purpose by nature. You tell them what you need, and they figure out the steps. The single-purpose, prescriptive integration model simply doesn’t work.
That’s where MCP (Model Context Protocol) comes into play. Instead of building specific integrations that do very specific things, MCP lets us expose a set of tools that any MCP client can use, in any way that you want to use them. All of your tools, all of your workflows, connected to Steady.
On top of the direct benefits you’ll get from Steady’s MCP server today, it’s also a concrete step toward realizing our vision for solving coordination in the agent era. An MCP server is crucial connective tissue for bridging the human / agent coordination gap.
What you can do with Steady’s MCP server
We could spend an entire post detailing the myriad workflows that a Steady MCP server opens up. For now, here’s a big-picture look at what you can do:
- Shift format - Live in a tool like Claude Desktop and prefer to read your digest there? Now you can do that.
- Conversational updates - Chat with Claude or ChatGPT to write and post check-ins and goal updates. Goal updates are a particular bright spot here; assistants will review past updates for the goal and help you close the loop.
- Chat with your Steady data - Echoes are designed for recurring context, not one-off questions. Until we build a first-class solution for one-off questions, this is a great way to meet the need.
- Connect data sources that don’t directly integrate with Steady - If you use tools that Steady doesn’t have dedicated integrations for, you can use MCP connectors in Claude and ChatGPT to assemble everything, craft an update, and push it to a check-in or goal update in Steady.
- Shift data from A to B - Want to automatically turn your Jira Epics into Goal Stories in Steady where they’re visible to the larger team? Now you can.
What’s live, what’s next
The server is already running in production behind a feature flag, on top of a full OAuth 2.1 implementation. The first release is focused on “core” data and workflows:
- Reading data from your Daily Digest, Smart Check-ins, Goal Stories, Goal Story updates, Echoes, activities, Insights, and absences.
- Creating Goal Stories and Goal Story updates.
- Checking in, end-to-end.
Future MCP server releases will focus on secondary tools like rendering Quick fill for check-ins and goal updates, Action Items, and more. We’d love to hear any specific needs you might have, or where the initial set of tools doesn’t meet your needs.
An important note on security: MCP connections in Steady are always personally scoped. In other words, you have access to the exact same data via MCP that you have in the Steady UI. Nothing more, nothing less.
Get on the beta list
If you’d like early access, join the beta list. We’ll pull people in waves over the next several weeks. Two asks from us: bug reports when you hit a snag, and feedback on how you’re using these tools in practice. We’d love to feature customer use cases in upcoming blog posts.