Case study
Erstatnings-Assistance.dk: An autonomous legal-tech agent
Architecting and building a complete, end-to-end AI product featuring a sophisticated RAG workflow, a stateful agent, and an "AI unit test" framework for reliability.
Simplifying complex legal research
Legal professionals and layman often spend hours sifting through vast document corpora to find precedents and build case strategies. Our goal was to create an autonomous AI agent for the legal-tech product www.Erstatnings-Assistance.dk that could dramatically reduce this time while discovering additional relevant precedents.
A stateful, autonomous RAG agent
We architected and built a fully autonomous AI agent that analyzes a 40,000-document corpus using a sophisticated Retrieval-Augmented Generation (RAG) workflow. This allows the agent to accurately reason upon information without costly fine-tuning. A key architectural feature is its ability to maintain a state of its search history to avoid redundant processing and logical loops.
The impact was immediate: a complex precedent search process that typically took legal professionals hours was reduced to mere minutes.
Ensuring reliability with "AI unit tests"
To ensure reliability for a non-deterministic system, we established a robust, containerized (Docker) MLOps workflow. This includes a full regression testing suite with "AI unit tests" that verify the agent's behavior against entire conversations. This directly addresses the critical business need for trustworthy and predictable AI, delivering a system where behavior is predictable and verifiable—a critical requirement for gaining customer trust and adoption.
Building for trust: Privacy-by-design
From its inception, the platform was architected for enterprise-grade security and data privacy. We implemented PostgreSQL's row-level security (RLS) to ensure robust GDPR compliance and the protection of highly sensitive user data. This ensured the product met stringent GDPR requirements from day one, demonstrating a commitment to responsible, secure AI systems—a critical differentiator in regulated industries.