ligi Lab: a short guide
ligi Lab looks at Maltese case law as a whole rather than one judgment at a time: how many judgments deal with a point, how a court tends to decide, and how things have changed over the years. This guide shows you how to use it, with a few searches worth trying first.
Open ligi Lab →Getting in
- Go to lab.ligi.ai. A laptop works best, although a phone is fine.
- You'll see a sign-in page from Cloudflare, which we use to keep the preview private. Its address starts with dark-butterfly-5279.cloudflareaccess.com, which is expected.
- Enter the email address we added for you, and you'll receive a one-time code by email. Type it in and the Lab opens. No password is ever asked for, and you stay signed in for 24 hours.
One search, four views
The bar at the top is your search. You narrow it with Words, date, court, party, judge, case type and More, and the line underneath tells you how many judgments match. The four tabs show that same set of judgments in different ways, so when you change the search, every tab follows.
As a sentence reads your search back in plain English, which is a quick way to check you've asked what you meant.
Words: type a term and press Enter or a comma, and it becomes a chip. A chip with several words is searched as an exact phrase, so unfair dismissal finds those two words together and in that order. A Maltese and an English term separated by a slash, such as danni morali / moral damages, finds either. With more than one chip, the All / Any switch decides whether a judgment must mention every chip or just one of them.
Results · find the judgments
The list runs newest first. Clicking a row opens a preview with an AI summary, the case details, Open judgment and a link to the judgment on eCourts.
Put danni morali / moral damages in Words, choose Civil Court, First Hall as the court, and set the date from 1 January 2020. Open a few judgments and see whether the summaries match what you'd expect.
Table · how does a court usually decide?
The Table counts the judgments in your search and breaks them down by year, outcome, court, case type and so on. You can switch between Counts and Share of row.
Choose the Rent Regulation Board as the court, open Table and count by Year, split by Outcome. Then click any number: the Lab opens exactly those judgments in Results and adds the matching filters to your search, marked "from Table". Back to the table returns you to where you were.
Compare · before and after, or one against another
Compare keeps your search and lets you add up to four scenarios, each with its own extra filters, so you can set periods, courts or case types side by side.
Choose the Rent Regulation Board as the court and Rent regulation old lease as the case type. Make scenario A date up to 31 May 2021 and scenario B date from 1 June 2021. "Claim partly upheld" goes from about 7% of 698 judgments to about 86% of 1,961, because since the 2018 and 2021 amendments most of these cases are rent reviews of old leases, in which the Board raises the rent but refuses the landlord's other requests.
Trend · is an argument gaining ground?
Trend shows, year by year, what share of the judgments in your search mention a term. Clicking a point opens those judgments.
- With no filters, add danni morali / moral damages: it rises from about 0.3% of judgments to about 6%.
- With the Rent Regulation Board as the court, add test tal-mezzi / means test: it is absent until 2018, around 14% in 2020 and 72 to 78% from 2022 to 2025.
Amounts · what is typically awarded
When your search includes Rent Regulation Board cases about a deposit, an Amounts card AI appears above the list. It shows the typical amount awarded and the typical deposit (the median), the range where half the amounts fall, and how many cases each figure is based on. Figures based on fewer than 20 cases are not shown.
So far we have read the amounts only in these deposit cases. We checked them by hand and with a practising lawyer, and other areas will follow.
Save and export
Save keeps your search together with the tab you're on, under Saved in the left menu. Export downloads what you're looking at as a spreadsheet.
Reading the numbers
- Anything marked AI (case type, outcome, amounts and the summaries) is generated by AI and still in beta, so please check it against the judgment before relying on it.
- Shares based on fewer than 20 judgments are hidden, and every share carries a 95% range.
- Individuals are never ranked or grouped, and in protected cases, such as criminal and family matters, names are not shown.
- The data currently runs to 30 October 2026, and newer judgments will follow.
- These are research findings from the case law, not legal advice.
Your privacy
During the preview the Lab records which pages you open, with your email address and the time, so we can see how it is used. Your searches are not recorded, apart from a short record when a figure involves a named person, which is a privacy safeguard. The record is kept on ligi's private server and deleted when the preview ends, within 90 days at most.
Tell us what you think
The Send feedback button at the top of the Lab opens an email with the page and your filters already filled in, so please use it whenever something strikes you, good or bad. We'd especially like to know:
- What was the first thing you tried, and did the Lab answer it?
- Was anything confusing, or did any wording make you stop and think?
- Which of the four views would you use in practice, and for what?
- Did any number look wrong to you?
- Which other amounts would you want the Lab to show, for example damages or awards in other courts?