Münch Energie makes project knowledge searchable with MünchGPT

Münch Energie develops renewable generation, grid infrastructure, and battery storage at industrial scale. Its Merseburg-Beuna battery cluster has 500 megawatt-hours of capacity. The company says it completes a new substation roughly every two months and draws on more than 20 years of energy infrastructure experience.

That pace creates a practical information problem. Every project adds files and knowledge that employees need to find, understand, and connect with the rest of the business. Münch Energie formed an internal AI team to improve those processes and built MünchGPT around the company’s own work.

The challenge: knowledge spread across company systems

Important project information sits in network folders and an internal wiki. Employees need to find the right file, understand its context, and connect it with information stored elsewhere. Folder structures and keyword search become less useful as the project base grows.

Münch Energie wanted one conversational entry point for this material. Its AI team also wanted to control the application and choose a suitable model for each task.

MünchGPT connects project files and the internal wiki

An MCP server makes the company’s network folders and internal wiki available to MünchGPT. Employees can ask questions in natural language and work with information from both sources through the same interface.

Münch Energie owns the product, its knowledge connections, and continued development. We provide the inference API behind the application. The internal team can focus on company processes and the quality of MünchGPT while using one integration for model access.

Several models through one inference API

MünchGPT currently uses Qwen3.6 35B, GLM-5.2, Qwen3-VL, and gpt-oss-120b. The model set gives the internal team options for language, reasoning, and multimodal work without tying the application to a single model.

Münch Energie selected LLMBase for the model choice and usage-based pricing. The usage-based plan matched the team’s demand better than a dedicated instance. Our OpenAI-compatible API keeps the integration the same as the team’s model requirements develop.

An internal product the team can keep extending

The responsibilities stay clear. Münch Energie’s AI team controls MünchGPT and the systems that contain company knowledge. We supply model access and metered inference through one API. Employees get a direct way to work with project folders and the internal wiki, and the team can add more processes and information sources over time.