Agentic Architecture for Medical Directory

Medifé, a leading private healthcare company in Argentina, sought to boost its business by incorporating Generative Artificial Intelligence (GenAI). Following a prior Proof of Concept (PoC), Ingenia was engaged to develop an AI assistant that would interact with members, allowing them to access medical directory services more efficiently through natural language.

The main challenge of the project lay in the definition and implementation of integration components (MCP) for agentic architectures.

Ingenia led the implementation of the AI Agent, a milestone achieved in several stages.

The initial phase defined the use cases, which involved converting the API "operations" for the medical directory and plans into "Tools" for the MCP. These tools were developed as functions encapsulating the necessary logic to consume the REST operations exposed by the APIs. Subsequently, an MCP server was configured, grouping all developed tools and exposing them under the MCP protocol. This server was deployed to Google Cloud Run, packaged as a Docker image, allowing it to run as a scalable service.

The MCP tools were then integrated into the assistant agent, enabling it to autonomously decide their use. Finally, a testing and refinement stage validated the assistant's behavior using the new tools exposed via MCP.

The result was a platform that allows the company to automate and streamline its directory services thanks to an AI agent. Key benefits of this platform include hyper-personalized experiences for over 300,000 members nationwide, in addition to achieving reduced operating costs, friction-less scalability, rapid time-to-market, and total interoperability between the agent and internal systems.