AI in Swiss law firms: where it saves time today, and how to start
Five areas where AI already saves time in a law firm, and a low-risk way to begin without putting the firm at risk.
Written with AI and reviewed by the Logol team.
Start with the work nobody bills for
Most conversations about AI in law begin with legal research or contract review. Those matter. But in most firms, the first real gains are found somewhere less prominent: in the administrative work that surrounds every matter. Time entries written on Friday afternoon from memory. Emails filed by hand. Conflict checks run in a spreadsheet. Monthly reports assembled by copying figures from three systems.
This work is repetitive, structured and easy to check. That makes it the right place for AI to start. The lawyer stays in charge of the legal judgement. The machine takes over the parts that follow a pattern.
Five areas where AI already saves time
1. Time recording
Time that is not recorded is time that is not billed. Lawyers know this, and still reconstruct their week from calendars and sent folders. AI changes the input, not the rule. A lawyer dictates or types a sentence such as “call with the client about the share purchase agreement, about 40 minutes”, and the system proposes the matter, the activity and the duration. The lawyer confirms or corrects. Entries are made closer to the work, with less effort and fewer gaps.
2. Intake and conflict checks
Opening a matter means collecting client data, identifying the parties and checking them against existing clients and opponents. AI can extract names and entities from an engagement email or a document, suggest matches including spelling variants, and rank potential conflicts by severity. The decision remains with the responsible partner. What changes is that a name spelled differently is less likely to be missed, and every check leaves a trace.
3. Document search
Every firm has the clause it drafted three years ago and cannot find. Keyword search fails when you do not remember the exact words. Natural-language search lets a lawyer ask for what they mean, for example a non-compete clause in a shareholders’ agreement, and get results across the matters they are authorised to see.
4. First drafts
AI is good at producing a structured first version: a cover letter, an engagement letter filled with matter data, a summary of a long email thread, a list of open points from a meeting note. It is not a substitute for legal drafting. It shortens the path to a draft that a lawyer then reviews, corrects and owns.
5. Reporting
Partners want to know where time goes, which matters are behind on billing and which invoices are overdue. When the data is held in one system, AI can help answer these questions in plain language and prepare the figures for the partner meeting, instead of someone assembling them by hand every month.
Why context matters more than the model
The same AI model gives very different results depending on what it can see. If time entries are kept in one tool, emails in another and documents on a file server, the AI sees fragments and has to guess. If matters, clients, emails, documents and time share one data model, the AI can propose the right matter for an email or a time entry, because it knows the context.
This is the practical lesson for firms evaluating tools. Ask less about which model is used. Ask more about which data the AI works with, and who controls it.
How to start with low risk
A careful start does not require a large programme. It requires a few clear decisions.
- Pick one workflow. Time recording and email filing are good candidates: frequent, measurable and easy to check.
- Keep a human in the loop. AI proposes, a lawyer or assistant confirms. No time entry, invoice or conflict decision should go out without a person approving it.
- Know where your data goes. Clarify where data is stored and processed, whether the AI provider may use it to train models, and how access is restricted. Swiss data protection law (the FADP) and professional secrecy under Art. 321 of the Swiss Criminal Code set the framework. Check the details for your firm with your own counsel or data protection officer.
- Define how success is measured. Before you start, note how long the task takes today and how often it goes wrong. Measure again after a few weeks, with your own numbers.
- Train the team, briefly and often. Short sessions on real matters work better than a single launch event.
- Write a simple policy. Which tools are approved, for which data, and whom to ask. Unapproved tools used with client data are a bigger risk than an approved tool used well.
Where ELLE fits
ELLE was built for this approach. It is an integrated CRM, ERP and DMS for law firms: one platform and one data model for clients, cases and conflict checks, for time, billing and accounting, and for documents and email. It runs in the firm's own Microsoft tenant, in its region. That is why its AI works with full context. The AI Time Tracker proposes case, activity and duration from a spoken or typed sentence. The Outlook add-in matches emails to the right case and learns from each correction. Conflict checks come with severity scoring and an audit trail. In each case, ELLE proposes and the lawyer confirms.
Whatever tool you choose, the principle stays the same. Start small, keep people in control, and measure the results with your own figures.