Companies need to integrate AI into their work processes, set clear rules and keep a human validation step.
In Switzerland, the issue goes beyond saving time. It is also necessary to protect data, respect professional secrecy and be able to verify the answers produced by the tool.
Why legal AI is becoming strategic for companies
Artificial intelligence is now used in many Swiss companies to analyse documents, search for information and draft texts.
For SMEs and large groups alike, the aim is no longer simply to test a tool. The point is to choose useful applications, measure the results and define the rules that govern its use.
The legal field is often a good starting point. AI can save time there, but it also requires a high level of security and traceability.
Silex presents its solutions by practice, in particular for in-house lawyers, fiduciary firms, banks and insurers. The partnership between Silex and EXPERTsuisse also concerns audit, tax and fiduciary professionals.
The most useful applications
Document analysis is one of the easiest applications to set up. AI can summarise large files, extract the key information and compare several versions of a contract.
It can also prepare draft memos, answer certain internal requests or produce a first summary before a legal professional steps in.
For risk management, AI can help to spot anomalies, check compliance with internal rules or prepare updates on regulatory developments.
Finally, it can speed up legal research and the analysis of complex contracts. Our article on AI for legal professionals shows, in particular, how to compare clauses with AI.
Law firms will find more specific applications in our article on AI for lawyers. For private individuals, our article on legal AI for individuals is more relevant.
Understanding the different types of AI
The term “artificial intelligence” covers several technologies.
Machine learning enables a system to identify patterns in large amounts of data. Deep learning uses more complex neural networks, notably for image or speech recognition.
Generative AI, for its part, is used to produce content: texts, summaries, analyses or code.
In a company, these technologies can complement each other. What matters is knowing what each tool is for, what data it uses and what risks its use creates.
AI-assisted legal analysis is precisely a way to integrate these tools without relieving the legal professional of responsibility for their decisions.
Choosing the technical environment
Choosing an AI solution is not just about the model used. Hosting, the deployment method and the way data flows also need to be examined.
A company can use a ready-to-use SaaS solution, develop its own system or rely on external cloud infrastructure.
Each approach offers a different level of control, cost and complexity.
For legal departments and regulated sectors, where data is hosted and how it is controlled often matter as much as the tool’s technical performance.
Complying with the FADP and, where necessary, the GDPR
In Switzerland, data processing that uses AI remains subject to the Federal Act on Data Protection (FADP).
Data protection must be taken into account from the design stage of the project, in accordance with Article 7 FADP.
The EU’s GDPR may also apply in certain situations, in particular when a company offers goods or services to people in the European Union. Our article on AI and data protection sets out these rules in more detail.
The company needs to know why the data is used, who the data subjects are and where the information flows.
If processing presents a high risk for the data subjects, a data protection impact assessment may be required under Article 22 FADP.
Access rights, encryption and the traceability of actions must also be defined before deployment.
Why the hosting location matters
Storing data in Switzerland can make sensitive files easier to manage and limit transfers to other jurisdictions.
The FADP allows personal data to be disclosed abroad under certain conditions, but companies need to know in which countries their information is processed.
According to Silex, its clients’ data is hosted in Switzerland by Infomaniak.
For legal teams handling confidential information, the hosting location, access rights and any use of files to train models must therefore be checked before any deployment.
The safeguards stated by Silex are detailed on the security and professional secrecy page.
Deploying AI step by step
An AI strategy should start with a specific business need, not with the purchase of a tool.
For example, the company may seek to speed up document research, improve the handling of internal requests or automate part of contract preparation.
It is then useful to test the tool on a limited project with a measurable result. The data used, the acceptable risks and the person responsible for validating the result must be defined in advance.
If the test delivers a real time saving without compromising quality or security, its use can then be gradually extended.
Team training remains essential. Users need to understand the tool’s limits and know that a generated answer must be checked before it is used.
Integrating AI into existing tools
A solution is more likely to be adopted if it integrates with the software teams already use.
The integration of Silex into WinLex and SkyLex, for example, makes it possible to start legal research without leaving the current file.
The partnership between Silex and MCR Solutions pursues the same goal: integrating AI-assisted research into existing work processes.
This approach limits changes to working habits and makes adoption easier for teams.
It also makes it possible to centralise the company’s legal knowledge more effectively and to avoid each employee working in an isolated tool.
General-purpose tool, in-house model or specialised solution?
A company can deploy an open-source model itself, use a general-purpose assistant or choose a specialised solution.
A model installed in-house offers more control, but requires technical skills and significant resources.
A general-purpose assistant is easier to use, but its data processing terms must be examined carefully before any sensitive information is sent to it.
A specialised solution such as Silex seeks to bring together legal research, case file analysis and data protection in an environment designed for Swiss law.
With Silo, teams can query their internal documents while linking them to the legal sources used in the analysis. The features are presented on the Silex product page.
According to Silex, client files are not used to train its models and each organisation’s data is kept separate.
The private bank Bordier & Cie has adopted Silex for its legal research.
Analysing complex files with verifiable sources
For a complex legal file, AI must be able to work on both the facts and the applicable law.
With Silo, Silex can use the documents uploaded by the organisation to understand the context of the file, then search for the relevant legal sources.
The answer can thus refer to an internal document, a statutory provision or a court decision.
The legal professional must then open the references, check the analysis and complete the reasoning where necessary.
The value of such a system therefore lies less in producing an automatic answer than in reducing the time needed to reach a verifiable analysis.
To test this approach on your own files, you can book a demo.
Costs
The price of a solution should be weighed against the number of users, the level of security, storage, the analysis features and the integrations available.
Silex offers a Solo plan at CHF 149 per month for one user, a Team plan at CHF 490 per month for up to five users and an Enterprise plan at CHF 895 per month for up to ten users.
Specific needs, such as certain integrations or dedicated support, can be covered by a tailored offer.
Details are available on the Silex pricing page.
In short
A successful AI strategy starts with a clear business need and a first measurable project.
Before deploying a tool, the company must check where the data is processed, whether it is used to train models and whether every important answer can be verified.
AI can speed up research, document analysis and drafting. The legal professional nevertheless remains responsible for validation and the final decision.
FAQ
Does AI replace lawyers and other legal professionals?
No. It can speed up certain tasks, but the results must be checked and decisions remain in human hands.
How can you tell whether data is sufficiently protected?
You need to check the hosting location, the transfer conditions, the access rights and whether files can be used to train the models.
How should a company start an AI project?
Start with a specific use case that is easy to measure, such as document research or contract analysis, then expand gradually if the test is conclusive.
