API to MCP

Inspiration We were inspired by the growing trend of conversational interfaces replacing traditional UIs. As users shift toward chat and voice, businesses will need to adapt fast—or get left behind. We asked ourselves: What if any API could become instantly usable via agents?

What it does Our tool turns any REST API into a fully functional MCP (Modular Computation Platform) server. This means you can plug your API into agent frameworks and let LLM-powered bots use it naturally through conversation—no extra frontend needed.

How we built it Since we believe all UI will step-by-step be replaced by chat agents, we think all companies will need to turn their APIs into MCP servers to provide the same services through chatbots and voice AIs. We created a set of agents that collaborate to:

Interpret OpenAPI specs or raw documentation

Autogenerate MCP-compliant modules

Validate and test them across real-world workflows

Challenges we ran into Parsing incomplete or outdated API docs

Ensuring deterministic function call outputs for LLM use

Harmonizing different agent roles in one seamless pipeline

Accomplishments that we're proud of Connected 5 different tools (OpenAPI parser, LangGraph, n8n, test suite, chat UI) into a single automated workflow

Deployed a working demo that lets any user convert their API into an MCP module in under 10 minutes

Made our agent system modular enough to be reused in other devops workflows

What we learned Building for agents requires a whole new design mindset—MCP is not just a format, it's a philosophy

Standard APIs rarely follow standards

LLMs are powerful, but combining them with rule-based validation drastically improves reliability

What's next for API to MCP Launching a self-serve interface so devs can upload their API and get a deployable MCP endpoint

Adding support for GraphQL and gRPC APIs

Partnering with agent tool providers to make MCP the default plug-and-play interface layer for AI agents

Built With

• amazon-web-services
• browserbase
• n8n