Open Energy Knowledge Gateway

A large amount of high-quality Swiss energy knowledge is publicly available in reports, publications and web content, but AI applications cannot easily access and reuse it in a structured and reliable way. Today, organisations often need to build their own retrieval systems and knowledge bases to make this information usable for AI assistants. The idea of this challenge is to create an Open Energy Knowledge Gateway that exposes curated energy knowledge through the open Model Context Protocol (MCP). A prepared Amazon Bedrock Knowledge Base will serve as the underlying knowledge source, while participants explore how to make its content accessible through an MCP server using AWS Bedrock AgentCore. The interface should allow different AI clients and applications to search the same knowledge base and retrieve relevant information together with its sources and metadata. During the hackathon, the team will build and test an end-to-end prototype and connect it to at least one external AI client. The broader goal is to explore how authoritative public energy knowledge could become reusable infrastructure for an ecosystem of AI applications.

The challenge could demonstrate how public institutions can move from simply publishing information for humans to also providing trusted knowledge for AI systems. We want to create easier access to authoritative Swiss energy knowledge, less duplication when developing AI and RAG applications, consistent access to sources and metadata, new opportunities for researchers, public authorities, companies and developers to build energy-related AI services, a reusable blueprint for exposing other public-sector knowledge bases through open standards.

A curated collection of publicly available Swiss energy information will be provided through an Amazon Bedrock Knowledge Base. Only information suitable for public use will be included in the hackathon knowledge base. Participants will not need to build the underlying knowledge base from scratch. Instead, they can focus on making the available knowledge reusable through the MCP interface and experimenting with applications that consume it.

Python, backend development Model Context Protocol (MCP) AWS / Amazon Bedrock / AgentCore, RAG, vector search or knowledge bases, LLM and AI application development. Most importantly, participants should be interested in exploring how open standards can make trusted public knowledge reusable by AI applications.