Frends MCP lets teams build, modify and troubleshoot integrations through Claude, ChatGPT and other AI clients
Frends is bringing enterprise integration work directly into the AI assistants that developers and business teams increasingly use, with the launch of Frends MCP, a Model Context Protocol-based capability designed to connect a customer’s Frends environment with AI clients such as Claude and ChatGPT.
The move addresses a persistent gap in enterprise AI adoption. While developers have rapidly adopted AI assistants for coding and productivity, integration development still typically requires switching between an AI client and an integration platform, manually configuring connectors and translating AI-generated suggestions into working processes.
Frends MCP is designed to eliminate that friction. Users can describe an integration requirement in natural language, after which the system searches the Frends Task catalog, selects and configures relevant connectors and produces a complete Process draft. Developers retain control by reviewing and approving changes within the Frends interface before anything is deployed.
“Every enterprise carries a backlog of integrations it has wanted to build for years but never had the capacity to do. Our customers can now put a request to their AI client and review a working draft within minutes, so teams can spend their time on the work that moves the business forward.” — Jukka Rautio, CEO, Frends
The capability also turns integration environments into a source of operational intelligence. Teams can query execution history to understand failed processes, identify slow-running integrations and map dependencies between systems reducing reliance on institutional knowledge held by individual employees.
Frends says internal testing showed a significant efficiency advantage from using purpose-built MCP tools. An AI client consuming raw data tables required approximately 10 million tokens for one query, compared with a few thousand tokens when the same question was handled through Frends MCP tools.
Security and governance remain central to the approach. Actions are tied to named users through personal tokens, with existing permissions determining what an AI client can access. Requests are logged, while drafts and changes require human review and approval before production deployment. The MCP server can also run in cloud, on-premises and air-gapped environments.
Frends MCP forms part of the company’s broader agentic strategy, alongside Frends Enterprise MCP, which exposes governed business processes as AI tools, and the Frends AI Connector, which enables AI-driven reasoning across multiple systems within a process.
The larger implication for enterprises is a shift from AI merely assisting developers to AI becoming an interface for the integration estate itself while governance and human approval remain in the loop.
