From idea to agent - a day full of AI at LM
It's early in the morning, the coffee machines are running at full speed and a day begins at LM on which AI is not discussed, but built. Colleagues from Development and Platform Services came together to pursue precisely this goal at the LM x AI Hackathon: AI solutions not only for our customers, but also to actively promote it internally at LM. We wanted to show that artificial intelligence can also create added value for us. It is a tool, an opportunity and a challenge at the same time. And above all, it can change our day-to-day work. So we asked ourselves: how can we turn existing use cases into real, functioning agents?
Practice instead of theory
The tasks were anything but abstract. We set ourselves real questions from our projects. Things that we deal with every day. With Copilot Studio, Microsoft Foundry (incl. Foundry Local) and Azure AI we structured ideas, connected data sources and built the first agents. Not as a quick demo, but as resilient approaches that can grow.
The use cases at a glance
Reuse Knowledge Agent
An AI agent who answers complex technical questions about infrastructure and technology projects. It is based on standards, existing documentation and empirical values from past projects. A particular focus was placed on the structured preparation of knowledge and the automated creation of documents.
Technologies used: Azure AI Search, Copilot Studio, Foundry, Power Automate
ISMS Agent
An agent to support information security and governance issues. In addition to answering queries, it automatically prepares ISO-compliant documentation so that important guidelines and evidence can be created in a structured manner. This creates a solid basis that simplifies and accelerates compliance work.
Technologies used: Microsoft Foundry, Copilot Studio, Azure AI
Partner Profiler Agent
This agent supports the structured research and evaluation of partner companies. Public data sources, partner information and digital profiles are bundled, analysed and prepared - as a sound basis for decision-making for further steps.
Technologies used: Microsoft Foundry, Copilot Studio, Azure AI
Local chat (Foundry Local)
Another exciting case was a local chat based on Microsoft Foundry Local. This agent enables secure access to company information directly on site, without a cloud connection, and provides quick answers to internal questions - ideal for data-sensitive scenarios.
Technologies used: Microsoft Foundry Local, Copilot Studio, Azure AI
Teamwork that works
What made the hackathon special? The collaboration. Mixed teams, open discussions, quick coordination and yes, pizza and caffeine as loyal companions. It was a day in which knowledge was shared, perspectives broadened and solutions developed together. This is exactly how we envision innovation.
What we take with us
The day clearly showed how much potential there is in the combination of expertise, curiosity and the right tools:
Several functional agents that are being further developed in a targeted manner.
Valuable insights into architecture, data sources and debugging.
Clear approaches for further use in projects, training courses and in the internal Centre of Excellence.
And now?
The first agents have been built, but this is just the beginning. In the coming weeks, we will continue to refine them and check which use cases can be scaled up. The hackathon has shown us that AI is not an end in itself. It becomes valuable when it makes our work easier, more efficient and smarter. A big thank you to everyone who took part - for their ideas, commitment and courage to try out new things. We are already looking forward to the next hackathon. Who knows what stories we'll be able to tell then?
Impressions of the AI Hackathon
Do you have any questions?
We support companies in this, Artificial intelligence in a practical and meaningful way - from the initial idea to the development of AI agents and integration into existing systems. With in-depth expertise and modern Microsoft technologies, we create solutions that simplify processes and deliver real added value. How can we support you in realising your AI projects?