In courts, document collections are constantly growing: pleadings, official notices, records of proceedings and archives of decisions built up over the years. With the electronic case file, this content is increasingly available in digital form – but this does not automatically make it findable or usable.
The IntraFind iFinder brings both worlds together in a single search: the case files with their pleadings, notices and documents, and the court’s internal archive of decisions. The latter, in particular, is often an underused source of valuable knowledge. Many decisions, particularly those concerning routine cases, are never published in commercial legal databases – within the court itself, they are the sole source for how comparable cases have been decided to date.
In addition to full-text search, iFinder analyses each document during the indexing process – that is, when the content is recorded and prepared for subsequent searches. The software automatically recognises certain characteristics and makes them available as structured metadata, for example:
This metadata can then be used to filter search results. No one needs to manually tag documents – judges can narrow down results to a specific case reference or legal provision with just a few clicks.
Generative AI builds on this foundation. The AI does not replace the search function, but works with its results: the IntraFind iAssistant uses the relevant files and judgements previously found via the search, formulates answers based on this information, and cites the underlying sources. The technical principle behind this is known as Retrieval Augmented Generation (RAG)
Using the ‘Ask a Document’ function in the iAssistant, judges can generate targeted summaries of selected case documents and ask questions to explore their content. The answer is based exclusively on these documents: what the proceedings are about, which facts are mentioned and what positions the parties are taking. Source references and quotations lead directly to the relevant section in the case file.
iAssistant searches the internal database of decisions for similar cases, comparable facts and existing lines of argument. The automatically recognised metadata improves the accuracy of the search – the quality of the AI’s response depends directly on the quality of the search.
The solution edits speech-to-text transcripts, corrects drafts, improves phrasing and provides working translations of individual passages – without the texts ever leaving the organisation.
Based on selected parts of the case file, the AI creates an initial draft of the facts of the statement of facts. AI-generated content serves only as a draft for review: The judge remains fully responsible for reviewing and assessing it.
For traditional legal research, courts use the AI services provided by legal publishers. However, case file contents cannot simply be entered into these systems. An internal AI environment addresses this challenge: it pseudonymises the case information and uses it to formulate an abstract legal research query. Only this pseudonymised query leaves the court; the link to the case remains internal.
All AI functions require the same foundation: an AI infrastructure in which sensitive and personal data are not transferred outside the court's IT environment. This includes operation within the court’s own data centre (on-premises), freely selectable language models and integration with the existing search system.
Courts do not need a general-purpose AI chatbot, but rather a solution that brings together search, internal knowledge bases and generative AI. IntraFind combines search within case files and judgements, intelligent search filters and AI assistance within a controlled, in-house environment. The key factor here is not just the language model, but the interplay between search quality, automatically generated metadata, access rights and data protection.
No. IntraFind’s AI solutions assist judges with research, the analysis of case files and the drafting of text. Judges remain fully responsible for reviewing and legally assessing the results, as well as for making the final judicial decision
Case file contents contain personal and confidential data. An in-house solution processes this data internally, takes access permissions into account and provides sources for its responses.
Case information is processed internally and pseudonymised. Only an abstract legal question is transmitted externally, without any reference to the specific case.
