Artificial intelligence is fundamentally transforming processes and best practices in internal investigations. The progress over traditional search term based approaches is now clearly measurable and close to revolutionary: AI delivers significantly greater speed, precision, completeness and cost control. With its proprietary AI solution, HM Argus, Hengeler Mueller is at the forefront of this development.
In several large-scale matters, HM Argus has largely automated manual reviews, strategically prioritized attorney analyses and drastically reduced time spent. The platform reliably processes large document volumes. Our experience shows that the currently higher initial effort for project setup, especially for prompt optimization, is more than offset by substantial efficiency gains during execution. Examples from our practice: In one review project, HM Argus reduced the documents requiring human review to a few thousand instead of many tens of thousands. In another case, the analysis of data equivalent to more than half a million binders could only be realized by the use of AI; a manual review would have required roughly 64 person-years.
Qualitatively, AI also outperforms search terms and human review. Thanks to contextual understanding, the AI identifies additional facts that would remain undiscovered search terms. At the same time, the false-positive rate decreases noticeably compared to keyword-based methods. Finally, the AI can process information across international projects in a wide range of languages and automatically translate it into the working language.
Why a proprietary solution? HM Argus is purpose-built for forensic inquiry and legal quality assurance: strict segregation of matter information, processing and storage within the EU, no use of project data for model training, full attorney confidentiality, and seamless integration into established eDiscovery workflows, including interfaces to the processing engine and review platform. The HM Argus platform consolidates modular components that can be deployed as needed: a chatbot for semantic and prompt-based searches, entity extraction for context-driven data condensation, and a reviewer for automated first level review and prioritized second level assessments. Specialized models can be integrated into the workflow wherever appropriate, such as for translations or named entity recognition. As a result, HM Argus is a precise toolkit for Early Case Assessment, the substitution of manual reviews, and more efficient fact development and processing.
Compared to the standard built-in AI features of major review platforms, we see three key advantages in this approach:
First: Adaptability and flexibility.
HM Argus can be finely tuned and continuously evolved based on project-specific requirements, question catalogs, factual segments and prompt libraries, independently of vendor limitations or platform lock-in.
Second: Integration capability.
RAG systems and vector indexes can be extended; internal and external knowledge bases (including client IP) can be incorporated in a legally compliant manner; and workflows can be iteratively scaled.
Third: Cost control.
The architecture enables pre-filtering, selection of the best-fit model, and tailored workflows per use case. This yields a strategic cost-benefit optimum that is already measurably reflected in comparison with alternative built-in tools.
This is particularly effective in our proven hybrid team setting. In matter-specific teams, attorneys, forensics and eDiscovery experts, IT specialists and data scientists work together seamlessly. This integrates legal assessment, technical forensics, and scalable data processing into an end-to-end workflow and allows us to continually adapt HM Argus to evolving project needs. It delivers a level of project-specific fit that a standardized tool cannot match.
