Make AI Agents Now Connect to Model Context Protocol Tools.

Make has added Model Context Protocol (MCP) support to its AI Agents, allowing users to connect external MCP servers directly in the agent builder. The update means agents can now access tools like HubSpot and Asana through pre-built connections or custom server URLs, all within the same workflow.

The integration addresses a gap between AI agent flexibility and automation scalability. Previously, users building agentic workflows had to choose between the adaptability of conversational AI and the structured reliability of automation platforms. With MCP support, Make is positioning itself to deliver both in a single scenario.

Announced on LinkedIn, the feature allows agents to invoke MCP tools during execution without leaving the Make environment. This follows Make's recent launch of Claude Skills and MCP Server support, which turned Anthropic's AI assistant into a Make automation builder.

Model Context Protocol, developed by Anthropic, provides a standardized way for AI systems to connect to external data sources and tools. In Make's implementation, users can add MCP connections through the agent builder interface.

The workflow is straightforward. Users select from pre-configured MCP servers for popular tools or paste a custom MCP server URL. The agent can then call those tools during the same run as other Make modules, whether that's querying a database, updating a CRM record, or triggering a notification.

This matters because it keeps the entire workflow visible and editable in one place. Teams don't need to manage separate agent configurations or API integrations; everything lives in the Make scenario builder where it can be monitored, versioned, and debugged alongside traditional automation steps.

The MCP addition arrives as Make continues expanding its AI capabilities. The platform recently added Claude Opus 4.8 for extended agentic workflows and launched MCP Toolboxes for permission control.

For teams already using Make, the value is immediate: AI agents can now access the same tools and data as their existing automations. A support agent might pull customer data from HubSpot, check project status in Asana, and draft a response, all without switching contexts or waiting for separate API calls.

The pre-built connections lower the barrier for non-technical users, while custom server URLs give developers flexibility to connect proprietary systems or niche tools. It's a practical middle ground between out-of-the-box simplicity and extensibility.

Competing platforms like Zapier and n8n have also been racing to integrate AI agents into their automation workflows, but Make's focus on MCP as an open protocol could give it an edge as more tools adopt the standard. The question is whether MCP gains enough traction to become the default way AI systems connect to external tools, or whether proprietary integrations continue to dominate.

Make's announcement included a blog post with additional documentation. The feature is available now to all Make users with access to AI Agents.