Swiss lawyers, judges, and compliance professionals have worked with legal databases like Swisslex and Weblaw for years. These platforms revolutionised legal research when they were introduced. Physical libraries were replaced by digital search. That was an enormous step forward.
But the requirements have evolved. The volume of data has exploded. Regulatory complexity has increased. Clients expect faster results. And the tools that were state of the art 15 years ago are reaching their limits.
What Traditional Legal Databases Offer
Traditional legal databases provide access to statutory texts, court decisions, and legal literature. Their core function: full-text search. The user enters search terms and receives a list of documents containing those terms.
This works. For experienced lawyers who know exactly which search terms are correct, which statutes might be relevant, and which court has jurisdiction. The problem: this type of research assumes the user already knows what they are looking for.
The strengths of traditional databases:
Extensive source collection. Platforms like Swisslex offer access to a large number of court decisions, statutes, and journals. The source collection has grown over years and includes material that would otherwise be difficult to access.
Reliable sources. The content comes from verified sources. When a court decision is in the database, it is authentic. For legal work, this reliability is fundamental.
Familiarity. Lawyers know these tools. They know how they work, which search strategies succeed, and how to interpret the results. This familiarity has a value that should not be underestimated.
Where Traditional Databases Reach Their Limits
The limitations become more apparent as requirements grow more complex.
Keyword Search vs. Semantic Search
Traditional databases work primarily with keyword search. Searching for “non-compete clause employment contract” returns documents containing exactly those words. Documents dealing with the same topic using different terms (such as “post-contractual restraint of trade” or “clause de non-concurrence”) are not found unless the user knows all relevant synonyms and variants.
AI-powered systems use semantic search. They understand the meaning of a query, not just the individual words. A search for “Can my employer prevent me from joining a competitor?” returns relevant results even if no document contains that exact wording. The system recognises that the question concerns non-compete clauses under OR Art. 340-340c and delivers the relevant statutes and decisions.
The difference is particularly relevant for lawyers researching outside their core area. A corporate lawyer needing to clarify an employment law question may not know all the technical terms. Semantic search helps even then.
Cross-Language Research
The Swiss legal system is trilingual. A federal statute exists in German, French, and Italian. Court decisions are written in the language of the proceedings. A Federal Supreme Court decision may be in German, French, or Italian.
In traditional databases, the user must search in each language separately. Researching a topic requires knowing the German, French, and Italian search terms and running three separate searches. Decisions in other languages are easily overlooked.
AI-powered systems with multilingual models can search across languages. A question in German also returns relevant French and Italian sources. The system understands that “Vertragsaufloesung,” “resiliation du contrat,” and “disdetta del contratto” refer to the same concept.
Cross-References and Context
Law does not exist in isolated documents. Statutes reference other statutes. Court decisions cite articles and prior decisions. Ordinances specify statutes in greater detail. FINMA circulars reference the Banking Act.
In traditional databases, these cross-references are partially available as links. But systematically navigating the web of relationships is cumbersome. Wanting to know which Federal Supreme Court decisions cite a particular OR article, how case law on that article has developed over time, and which cantonal courts have ruled differently requires dozens of individual queries.
AI systems with citation graphs map these relationships in a structured way. The user sees at a glance which sources are linked. Which decisions cite an article, which articles are referenced in a decision, how a line of case law has developed.
Synthesis Instead of Listing
Traditional databases deliver lists of results. The user receives 50, 200, or 500 hits and must determine for themselves which are relevant. Reading, filtering, and summarising the results is the actual work.
AI-powered systems can deliver a synthesis. Instead of a list of 200 decisions, the user receives a structured summary: the leading decisions, the essential arguments, the development of case law, and references to the original sources for verification.
The lawyer saves time not only in searching but above all in evaluating the results.
Cantonal Coverage
Cantonal law is often incompletely covered in traditional databases. Federal law is well indexed. But cantonal procedural codes, municipal regulations, and cantonal court decisions are frequently only partially available or not available at all.
For lawyers working across cantons, this is a significant obstacle. They must visit cantonal portals individually, navigate different user interfaces, and compile information manually.
An AI system that aggregates all 26 cantonal legal systems and makes them accessible through a unified interface solves this problem.
What AI-Powered Knowledge Systems Do Differently
The fundamental difference: traditional databases are document collections with search functionality. AI-powered knowledge systems are structured knowledge bases with comprehension capability.
Retrieval-Augmented Generation (RAG). AI systems combine structured databases with language models. The language model understands the user’s question. The retrieval system finds the relevant sources. The result: a coherent answer with source references instead of an unsorted list of hits.
Vector search with BM25 hybrid. Modern systems combine semantic search (vector search) with classical keyword search (BM25) and reranking. This yields higher hit accuracy than any single method alone.
Structured units instead of full-text documents. Instead of indexing entire documents, statutes are broken down into their components: articles, paragraphs, sub-clauses. Decisions are structured into considerations, facts, and rulings. This granularity enables more precise results.
Citation graph. The relationships between documents are modelled as a graph. Which decision cites which article? Which articles reference each other? Which decisions form a line of case law? These relationships are automatically extracted and made navigable.
The Transition
Switching from a traditional database to an AI-powered system is not an either/or proposition. Many firms will use both tools in parallel, at least during a transition period.
What changes is the nature of research. Instead of beginning with keywords and manually filtering results, the lawyer starts with a question in natural language and receives a structured answer with verifiable sources.
The lawyer remains responsible for the analysis, the argumentation, and the judgment. But the information gathering that previously consumed the bulk of research time is drastically accelerated.
Enclava: AI-Powered Legal Database for Switzerland
The Enclava platform combines all of these advantages in a system purpose-built for Swiss law: 27,795 statutes (federal and cantonal), over 1.1 million court decisions, 2 million structured legal units, 1.4 million citation graph edges. Semantic search in German, French, Italian, and English. Complete source references to Fedlex and cantonal portals. Swiss hosting, Swiss jurisdiction.
For legal professionals ready to take their research to the next level: visit enclava.ch or contact us at [email protected].