🚀 Product Update: Dume.ai Now Supports MCP Integration

We're excited to announce that Dume.ai now supports MCP (Model Context Protocol) — a powerful new way to bring external, real-time data into your AI workflows.

Whether you're building complex workflows, automating decision-making, or just want your AI agents to operate with up-to-the-minute context, Dume.ai MCP integration opens the door to a smarter, more connected system.

🔍 What is MCP and Why Use It?

MCP (Model Context Protocol) is an open standard that makes it easy for AI systems to communicate with external tools, APIs, and services in a structured way.

With MCP, your AI agents can:

Fetch real-time data from internal or third-party systems
Trigger actions or pull state changes during a conversation
Automate workflows across multiple platforms

In short, MCP bridges your tools and your AI logic, enabling true external data workflow automation.

🤝 Official MCP Integration Powered by Composio.dev

We've partnered with Composio.dev to make it easy to get started with a managed MCP server. However, you're also free to use your own MCP-compliant server — whether using SSE or HTTP-streamable endpoints.

🛠️ How to Use MCP with Dume.ai

Here's a step-by-step guide to connect an MCP server to Dume.ai — using Composio's MCP server or your own.

🔧 Using Composio as Your MCP Provider

Create an Account Go to mcp.composio.dev and sign up or log in.
Authenticate Follow the prompts to authenticate and activate your MCP server.
Name Your Server Give your server a name and click Next.
Select Actions Choose the actions/tools you want your MCP server to support.
Create the Server Click Create Server to generate your instance.
Choose Client Type In the dialog that appears, select HTTP-Stream as the MCP client type and copy the generated URL.

🧩 Add MCP Server in Dume.ai

Login to Dume.ai Go to app.dume.ai and log into your dashboard.
Navigate to MCP Configuration Find Settings > MCP Configuration and click Add New Server.
Add the Server Choose the HTTP-Streamable type, enter the name of the server, paste the URL from Composio, optionally add a description, and click Add Server.
Test the Connection You'll now see the server listed with a "Disconnected" status. Click Test to verify the connection.
Connection Confirmed If everything's working, the status will change to Connected.

✅ That's it — your MCP server is now ready to be used in Dume.ai chats and workflows!

🌐 Bringing Your Own MCP Server?

Not using Composio? No problem.

Dume.ai supports both:

HTTP-streamable endpoints
SSE (Server-Sent Events) based custom MCP servers

As long as your server follows the MCP specification and exposes a valid endpoint, you can add it using the same steps — just select the appropriate connection type during setup.

🧠 Tips & Troubleshooting

Test First: Always use the "Test" button to validate your MCP server setup before relying on it in production workflows.
Auth Errors?: Ensure your tokens or keys are valid and headers are properly set.
Stream Issues?: Double-check your server supports HTTP-stream or SSE as expected.

🔄 Use Cases for MCP + Dume.ai

Once your MCP server is connected, you can:

Inject external data live into chat prompts
Trigger workflows based on external events or metrics
Build adaptive workflows that pull in latest status from tools like GitHub, Notion, Google Sheets, and more

This is true external data workflow automation, powered by LLMs that understand and act on real-time data.

🚀 Try the Dume.ai MCP Integration Today

Ready to connect your world? Whether you're exploring how to use MCP with Dume.ai, looking to connect an MCP server to Dume.ai, or curious about external data automation, this is your next step forward.

👉 Start Building with MCP on Dume.ai

Have questions? Join our Developer Community or reach out to support.