The Model Context Protocol (MCP) is the new gold standard in the Artificial Intelligence industry. Developed to solve the isolation of Large Language Models (LLMs), MCP acts as a "universal connector" that allows AI to access external data and tools instantly and securely.
At Pipefy, we’ve adopted MCP to transform your AI Agents 2.0 from simple consultants into real executors. With this protocol, your agents are no longer limited to what is inside the platform; they gain the "hands" needed to interact with your entire software ecosystem, including Slack, Jira, Salesforce, Google Drive, and more.
What MCP solves
End of manual integrations: Instead of building complex APIs for every tool, MCP provides a single language for the agent to communicate with any system.
Real-time context intelligence: Your agents can query external databases in real-time before making a decision or filling out a field.
Implementation agility: Developers and IT teams can connect their own MCP servers to Pipefy in minutes.
Strategic benefits
Universal Orchestration: Pipefy acts as the brain (Control Plane), while MCP provides the connection (Execution Plane) to the external world.
Enterprise-grade security: All data exchanges via MCP respect Pipefy’s governance layers, ensuring the AI only accesses what it is authorized to see.
Scalability: As an open and rapidly expanding standard, MCP gives your operation the flexibility to swap or add tools without having to rebuild your automation flows.
Practical examples in your workflow
With MCP, you can configure two types of skills for your agents:
Actions: The agent performs a task outside of Pipefy.
Example: "After a reimbursement is approved, automatically create the payment in SAP or send the receipt via Slack."
Data Retrieval: The agent searches for information to enrich the process.
Example: "Whenever a new support card arrives, consult the customer's history in HubSpot and summarize the account status for the analyst."