🤖 AI Agents in Action: Automated Support for Document Imports in the BA’s E-File System
At the German Federal Employment Agency (Bundesagentur für Arbeit, BA), importing documents into the electronic file system (E-Akte) relies on a modern, cloud-native architecture using technologies such as Quarkus, Apache Kafka, and S3. Communication follows an asynchronous request-reply pattern, where every sender receives a return receipt confirming the import.

Despite the system’s robustness, common support queries persist — like “Where’s the receipt for my documents?” or “Can you resend the confirmation?”
Answering these typically involves manual investigation across multiple systems and interpreting complex data scenarios.

To address this, the BA developed a proof of concept (PoC) for an AI-powered agent that automatically analyzes relevant data and generates clear, accurate responses for users.
The result: a significant reduction in repetitive manual tasks and a boost in operational efficiency.

🧠 This talk will cover:
The real-world use case and its complexity

Key challenges and design decisions in building an AI agent within a public-sector environment

Practical insights and takeaways from the PoC implementation

Ideal for those interested in AI integration, automation in public services, or real-world agent-based applications, this session offers hands-on knowledge from a truly impactful project.