Subscribe to our newsletter

Blog

How to use AI to analyse and compare clauses in Switzerland

May 28, 2026

AI can extract, summarise and compare clauses much faster. To remain reliable in Swiss law, however, it must keep the sources, protect the data and leave the final decision to the lawyer.

The aim is not to replace human analysis. AI speeds up the identification of information and brings out important differences, while the professional checks their legal significance.

‍

The limits of a general-purpose AI when faced with a clause

A legal clause cannot be treated as a simple text to be summarised. Its scope depends on the contract as a whole, the applicable law, its date and the other provisions of the document.

A general-purpose assistant can produce a clear but incorrect answer, especially if it is not based on verifiable sources.

In a legal file, the tool should be used to spot obligations, differences or sensitive points, without replacing the lawyer’s interpretation.

Individuals face the same risk when they use a general-purpose AI. Our guide to legal AI for individuals sets out these precautions.

‍

A reliable comparison must remain verifiable

A legal analysis is only really useful if the professional can retrieve the sources on which it is based.

When a tool flags that a clause presents a risk or that a rule applies, the lawyer must be able to check the statutory text, the case law, the legal literature or the internal document used.

This traceability makes it possible to distinguish genuine legal research from mere text generation.

Our article on AI-assisted legal analysis sets out this method in detail.

‍

Protecting professional secrecy and data

The contracts and files analysed by lawyers often contain confidential information.

Before sending a document to an AI, it is therefore necessary to know where the data is stored, who can access it and how long it is retained.

The guidelines adopted by the Swiss Bar Association in June 2024 stress the importance of these checks for professionals bound by secrecy.

The firm’s documents should not be used to train general models either.

The applicable rules are set out in our articles AI for lawyers and AI and data protection.

‍

Summarising long documents without losing important details

Automatic summarising is particularly useful in large files, due diligence exercises or contract reviews.

AI can bring out the main obligations, time limits, conditions and sensitive clauses without requiring the lawyer to carry out a full first reading of hundreds of pages.

The summary must never, however, cut off access to the original text. A nuance, an exception or a condition can change the meaning of a clause completely.

The lawyer must therefore be able to return immediately to the relevant passage to check the analysis.

‍

Comparing several clauses and spotting the differences

AI can compare two versions of a contract or compare a draft with an internal template.

It can flag a change that looks minor in the wording but actually alters the allocation of risk, for example in a liability or force majeure clause.

It can also detect a contradiction between the main contract and its annexes.

To be useful, the comparison must go beyond simply spotting the words that have changed. It must bring out the differences that may have a legal effect, then leave the professional to assess their importance.

‍

Using internal documents with Silo

The analysis becomes more relevant when the tool can also take into account the documents already used by the firm.

With Silo, Silex makes it possible to work from templates, precedents and internal documents in a separate environment.

The lawyer can then compare a new clause with the standards already approved by the organisation or retrieve a position taken in an earlier file.

This approach helps keep drafting consistent across several files without having to search the archives by hand.

Silo’s features are presented on the Silex product page.

‍

The lawyer keeps the final decision

AI can identify a risk, but it does not know whether that risk is acceptable for a given client.

It does not know the full negotiation strategy, the commercial objectives or the compromises the client is prepared to accept.

The tool’s role is to prepare the analysis and bring out the points that deserve attention.

The lawyer retains responsibility for the interpretation, the advice and the final decision.

‍

Training teams to use AI reasonably

Using AI also requires new habits.

Professionals must learn to ask precise questions, check the sources and spot answers that look convincing but remain fragile.

Training should also cover confidentiality and the internal rules on which documents may be sent to the tool.

The aim is not to delegate everything to AI, but to use it on the tasks where it really adds value: research, comparison or document synthesis.

‍

Moving from a test to organised use

In a large firm or a legal department, the use of AI cannot rest solely on each colleague’s individual practices.

It is necessary to define which documents may be analysed, which tools are authorised, who may access them and how the results must be checked.

Our guide to enterprise AI strategy sets out this approach in more detail.

Suitable hosting and clear rules on data remain essential when the tool is used at scale.

‍

Silex in the tools professionals already use

Integration into existing software makes AI easier to adopt.

The partnership between Silex and MCR Solutions aims to integrate AI-assisted legal research into everyday work processes.

With Silex in WinLex and SkyLex, lawyers and notaries can also query the tool from their file.

The private bank Bordier & Cie uses Silex for its legal research, while the partnership between Silex and EXPERTsuisse concerns in particular audit, tax and fiduciary professionals.

‍

Saving time with Silex without losing control

Silex combines sourced legal research with analysis of the documents on file.

According to Silex, data is hosted in Switzerland and client files are not used to train the models. The safeguards are detailed on the security and professional secrecy page.

To summarise a document, compare clauses or analyse a file, the tool provides results linked to their sources so that the lawyer can check them before using them.

The Solo, Team and Enterprise plans are priced at CHF 149, CHF 490 and CHF 895 per month excluding VAT. Details are available on the Silex pricing page.

The solutions by practice page presents the uses according to each profession. To see the tool on a real file, you can book a demo.

‍

Before using AI to analyse clauses

Before working on a real contract, check four points above all: where the data is hosted, whether it is used to train the models, whether the legal sources can be verified and whether the original text remains easy to access.

A useful comparison must also bring out the legal meaning of the changes, and not only the words that have changed.

‍

In short

AI can speed up the reading, summarising and comparison of contracts.

It is particularly useful for spotting differences, retrieving information and preparing a first analysis.

Data confidentiality, source-checking and the final interpretation nonetheless remain the lawyer’s responsibility.

‍

FAQ

Does Silex store the documents analysed?

According to Silex, documents placed in Silo are kept in Switzerland in a secure environment and are not used to train the models.

Can AI be used to draft a complete contract?

It can help prepare or structure a draft, but the lawyer must review, adapt and validate the document before any use or signature.

Can the tool be used with documents covered by professional secrecy?

That depends on the safeguards offered by the solution. Silex states that it was designed for the requirements of Swiss professional secrecy, with hosting in Switzerland and data-protection measures.

‍