Anyone who thought the legal sector was the last bastion against the rise of AI is mistaken. Three-quarters of German companies will be using AI in some form by 2025—and in the legal sector, adoption is already surprisingly high. 56 percent of legal teams use generative AI, and 66 percent have a dedicated legal tech budget. These are no longer predictions for the future; this is the current state of affairs. The industry will reach a true turning point in 2026—and anyone who hesitates now risks falling behind.
Where AI Really Lands in Legal Practice
According to recent surveys, the three areas with the greatest potential for AI in everyday legal work are document analysis (22%), legal research (21%), and document drafting (20%). That sounds technical, but in practice it means: hours turn into minutes.
Document analysis is perhaps the most obvious area of application. AI systems scan through contracts, pleadings, and court decisions in seconds, highlight critical clauses, identify contradictions, and summarize key points. What used to keep a paralegal or junior attorney busy for days is now delivered by the machine as a structured report—for review, not for blind adoption.
AI-assisted legal research fundamentally changes the relationship between lawyers and databases. Instead of manually searching through commentaries and case law databases, systems like juris with AI search or Beck-Noxtua enable semantic search: You ask the question in natural language, and the AI delivers relevant results—including an assessment of their relevance. This is no longer just a search engine; it is a research assistant.
Document creation closes the loop: Based on templates, client data, and research output, tools such as Justin Legal, JUPUS, or JUNE generate draft pleadings, complaints, or contract clauses. The lawyer sets the parameters; the AI produces the initial draft.
The market for legal tech tools — an overview
The German legal tech ecosystem has become more dense than many realize. In addition to established providers, numerous specialized solutions have emerged, targeting different types of law firms and legal departments.
In the research and knowledge sector, well-known publishers with AI-powered platforms dominate: Beck-Noxtua (C.H. Beck), juris AI Research and Wolters Kluwer Online offer AI integration into existing databases. The advantage: the data is verified, significantly reducing the risk of hallucination in source citations.
For law firms looking to automate their processes, there is a growing range of specialized tools: ANITA and Libra.ai for contract analysis, BEAMON AI for structured document processing, Hello Kira for client-focused communication support. MAIK (by Methodigy) specifically addresses the workflows in mid-sized law firms. Prime Legal AI and JURA AI Assistant (integrated into RA-MICRO, one of Germany’s most widely used law firm management systems) significantly lower the barrier to entry for smaller firms.
What’s missing from this list? A clear market leader. This is both an opportunity and a challenge—the landscape is fragmented, interoperability is often lacking, and choosing the right tool requires a genuine examination of one’s own workflows.
Three tools for mid-sized law firms: First juris AI Research — because the database is verified and errors in source citations are structurally minimized. Second JURA AI Assistant in RA-MICRO for law firms already working on this platform — seamless integration. Third JUPUS or Justin Legal for document creation, where the ROI is felt the fastest. Everything else is still experimental at this point.
AI in the judiciary: Not just a matter for law firms
Digitalization does not stop at the courts—albeit at a slower pace than in the private sector. A concrete example from Germany: The Lower Saxony Ministry of Justice put the AI assistant EMIL into productive operation. EMIL supports judges in researching and preparing decisions. It is an early but significant signal: even government institutions are no longer testing AI only in the lab, but in real-world operations.
The reform of the monetary limits for civil litigation in Germany (effective January 1, 2026) has an interesting side effect: it intensifies the use of AI in civil procedure law. More cases that fell below the old limits now potentially end up in court—and this increases the pressure to make proceedings more efficient. AI support for case file analysis and judgment preparation becomes more attractive as the volume of cases the courts must handle increases.
Risks no lawyer can afford to ignore
Anyone who believes AI tools are plug-and-play will sooner or later find themselves in an embarrassing situation. The case before the Cologne District Court is a prime example: A lawyer submitted an AI-generated brief—including citations and rulings that simply do not exist. In AI jargon, this is called “hallucinations”; in the courtroom, it means embarrassment and potential professional consequences. Similar cases in the U.S.—involving fabricated Supreme Court rulings—have put the issue on the international agenda.
The professional duty to critically review work is not a new concept; it applies to AI output just as much as it does to research conducted by an intern. The Anwaltsblatt puts it aptly: Uncontrolled dissemination of AI-generated documents may constitute a breach of the retainer agreement. This means that anyone using AI must possess three competencies—understanding the basic principles of the systems used, critically evaluating the output, and formulating prompts in such a way that the result is usable.
AI competence has thus become a core competency for lawyers — not optional, but part of the professional toolkit.
Regulation: The EU AI Act and Switzerland are following suit
The legal framework for AI systems is undergoing a transformation—which has direct relevance for everyone who uses AI in the legal field. The EU AI Act is taking full effect gradually: General-purpose AI models have been regulated since August 2, 2025, and broad applicability across most sectors takes effect on August 2, 2026. For law firms and legal departments, this means that the systems they use are subject to transparency and risk management obligations—and potentially also their own processes in which AI plays a decision-making role.
Switzerland is not an EU member—the EU AI Act does not apply directly here. This is a significant difference: Swiss law firms are currently not subject to AI regulatory obligations as strict as those of their counterparts in Germany or Austria. However, this window is closing: The Federal Council has tasked the FDJP with drafting legislation by the end of 2026 to implement the Council of Europe’s AI Convention—a consultation draft is currently in the works. For Swiss law firms, this means: Now is the time to establish processes before requirements become more stringent.
What’s interesting here is that the law regulates AI, and at the same time, AI helps us understand the law. This dual role makes the industry one of the most exciting areas to watch in the entire AI ecosystem.
Conclusion
AI in the legal field is no longer just hype—it is an operational reality. The tools exist, acceptance is growing, and the competitive pressure on law firms and legal departments to work more efficiently is real. At the same time, the technology is a tool, not a substitute for legal judgment. Anyone who uses AI uncritically risks more than just an embarrassing moment in court—they risk the very foundation of their client relationship.
The way forward does not lie in blind adoption, nor in outright rejection. It lies in informed, competent use—with an understanding of what the systems can do, what they cannot do, and what remains the responsibility of humans in every case: accountability for the outcome.
🎼
The Doctor’s Opinion
It took the legal profession decades to accept simple word processing—now AI is supposed to transform the profession in just a few years. I consider this realistic, but the industry systematically underestimates how much expertise this requires: not only technical, but also ethical and in terms of liability law. A delusional judgment in a brief is not a software error—it is a human error that should have human consequences.
