Pandektes, a Copenhagen company that builds a searchable database of European law, has raised €13.5M in a Series A led by the Munich fund Alstin Capital. The four-year-old firm will use the money to cover more jurisdictions and to open the database to other software companies.
PROfounders and Scale Capital joined the round alongside existing shareholders, the Copenhagen company said on 5 October, having raised about €3.9M before this, including a €2.9M seed in February 2025.
The product gathers legislation, court decisions and administrative rulings, structures them so they can be searched together, and connects them to a customer's own internal documents. A lawyer can run a question across more than one country rather than working through a separate national database for each.
The second half of the strategy is what the round pays for. Pandektes is opening that organised data to other legal software companies through a direct feed, so they can build their products on it instead of assembling the raw law themselves.
Its own numbers describe a company early in that transition. More than 500 law firms, companies and public institutions use it, revenue grew six-fold over twelve months, and about two thirds of trial users convert to paying, from a base of roughly 55 staff. Denmark, Germany and Switzerland are the markets covered today, the Danish grid operator Energinet is among the named customers, and the company intends to extend across Europe and into North America.
Casper Laursen founded Pandektes in 2022 with Søren Brokær and runs it as chief executive. "Building a serious challenger was always deemed impossible," he said. "Today, we can collect and make sense of legal information at a scale that was not feasible before."
Andreas Schenk of Alstin Capital described the opening the fund saw. "We have rarely seen a market like legal research, where so many end-users would like to switch their core system but historically simply couldn't."
The Same Owners For Decades
The market Pandektes is entering is one of the most consolidated in professional software. LexisNexis, owned by RELX, and Westlaw, owned by Thomson Reuters, dominate the English-speaking world, while C.H. Beck holds a comparable position in Germany and Karnov Group does across the Nordics.
What protected them was collection cost rather than technology. Someone had to digitise every ruling, write the summaries, maintain the citation links that show whether a judgment still stands, and repeat that country by country in each national language.
The prices those businesses command show what the work is worth. When Karnov bought the Spanish and French legal information arms of Thomson Reuters and Wolters Kluwer in 2021, it paid around €160M for units turning over €123.9M between them, and it sized the addressable legal information markets it was buying into at about €413M in Spain and €1.05B in France. Those are two countries, and the pattern repeats across the continent.
That history is the context for Laursen's remark about a challenger being deemed impossible. The argument now is that the collection work costs a fraction of what it did, because machine reading can structure a corpus that previously needed editorial staff.
The Money Went One Layer Up
While Pandektes was raising €13.5M, the application companies above it were raising considerably more. Harvey raised $550M in September at a $15.6B valuation, with annual recurring revenue above $400M and more than 3,000 customer organisations, among them 80% of the Am Law 100, while its Swedish rival Legora has been reported seeking money above $10B.
Those platforms draft, review and reason over documents. What they do not own is the primary law underneath, which in Europe sits in separate national systems, in different languages, under different publication rules.
Pandektes is positioning in that gap, which is a reasonable place to stand and an exposed one. Selling to the application layer means the customers are well funded, and it also means they can build or buy the same data if it becomes central enough to their product, which is the same bind faced by anyone opening their data to AI tools. The company is also competing with the incumbents on their own ground, since LexisNexis, Westlaw and Karnov already sell to law firms directly and have been attaching their own AI products to the databases they own.
Grounding Reduces Errors, Not To Zero
The evidence for why the data layer matters comes from a study that also sets its limits. Researchers at Stanford tested commercial legal research tools against 202 questions in 2024, with two lawyers reviewing every answer blind, and found that Lexis+ AI produced incorrect or unsupported responses in roughly 17% of cases while a general-purpose model with no legal database behind it did so in 88%. Grounding a model in authoritative law therefore made a substantial difference, which is the commercial argument for what Pandektes sells.
The same study found that it did not make the tools reliable. Westlaw's AI-assisted research erred in about a third of answers in the expanded version of the work, and the researchers concluded that retrieval from a legal database reduces hallucination without eliminating it.
Thomson Reuters disputed the methodology, saying the first round had tested a narrower product than its main research tool, and Stanford later extended the study after being given access. The dispute itself illustrates how much rests on the quality of the retrieval layer rather than the model.
Each New Country Is A New Build
Three jurisdictions is a narrow footprint for a company describing itself as European legal infrastructure, and widening it is the work the round pays for. A new market means a different court system, a different publication regime for judgments, a different language and usually an entrenched national publisher with a long-standing relationship to the bar, which is the combination that kept the market closed, and automation reduces the cost without removing the requirement.
North America, the other stated destination, is where Westlaw and LexisNexis are strongest and where the best-funded application companies already sell. Six-fold revenue growth across a year is a strong figure from a small base, and the two-thirds trial conversion rate suggests the product answers something customers were looking for, though neither tells an outside reader what the business turns over.
What this round establishes is that a European investor is willing to fund an attempt at the layer underneath legal AI, on the view that the data has become collectable by a company of 55 people. Whether it travels beyond Denmark, Germany and Switzerland is the part the next two years will show.
