Let me tell you a story about a side project that turned into a full-blown legal tech revolution.
It all started with a simple idea: “Let’s build a search engine for legal documents — something like Google, but for court decisions and regulations.” That’s what Filip Boltužić and his team set out to do.
Sounds harmless, right? Just index a bunch of dry PDFs.
But then came the plot twist — AI entered the chat.
From Boring Indexing to Brainy AI
Filip’s no stranger to building tech. With a background in engineering and machine learning, he figured they could just scrape public data and slap on a basic search tool.
Easy win. Except… the public legal data landscape in Croatia was a dumpster fire.
“For example, only around 3% of court decisions were actually publicly available,” Filip told me.
“And even that was a mess.” We looked at each other and laughed, as I know very well what he is talking about.
So they scraped what they could, hit up institutions, pulled what they could from APIs, and stitched together a Frankenstein of legal documents. Still, they were trying to solve the wrong problem.
Then AI changed the game. Not only could it index documents — it could understand them. It could perform logical searches. Connect references. Compare rulings. All in milliseconds.
Boom. Ulpian was born.
The Wikipedia of Legal Documents… But on Steroids
Imagine asking, “Which Croatian court rulings from 2021 support the right to compensation in public procurement disputes under EU law?”
A junior lawyer would spend two days digging, sweating, and cursing their life (this is just my guess; I have nothing against honest day work in the legal trenches).
Ulpian does it in milliseconds.
This isn’t just a fancy document viewer. It’s like Wikipedia got a law degree and a Red Bull addiction in the same basket.
Every document is interlinked. Citations point to actual texts. References click through to sources. The whole network is traversable like a knowledge graph, but faster than any human could ever manage.
And the real kicker?
“It’s a stack of AI models working together,” Filip explains.
“One interprets the query, another finds the right legal context, another pulls the most relevant data.” Details are part of their secret sauce. And the sauce is spicy!
Oh — and they’re not just stopping at Croatia. Ulpian is already testing in Austria and Germany, where (surprise!) the legal doc chaos is just as real. The scalability of this thing is huge, and Filip estimates they could be operating in the whole EU within the next 5 years. And that is the power of AI for you!
Will Junior Lawyers Be Replaced?
I had to ask: Does this mean fewer jobs for junior associates?
Filip shook his head.
“No — they’ll just be able to do more. Faster. Better. The tool doesn’t replace judgment — it just eliminates the grunt work.”
So instead of spending days building a reference list, juniors can focus on strategic thinking. And firms? They can handle more cases without burning out the team.
Seems like a win-win. Except for lawyers who secretly enjoyed Ctrl+F-ing 900-page PDFs. If any are reading this, you guys are toast…
I would say that lawyers who ignore AI aren’t just behind — they’re billing hours in the Stone Age.
Already Making Money
Unlike most early-stage AI startups, Ulpian isn’t just hype. They’re already bringing in revenue.
Their clients? Mainly legal firms and corporate legal departments who are tired of wasting time and money on inefficient research.
“It’s like ChatGPT, if ChatGPT passed the bar exam in five languages.”
The most common use case? Building lists of reference cases with legal precedent — something that used to take days.
Now? Seconds.
This is why Ulpian might just become the “Slack for legal research” — something every law office needs, whether they like it or not. The brutal truth is it will be “Cherish or perish!”.
And Then He Dropped a Killer Startup Idea
At the end of the interview, I asked Filip for a nugget of wisdom — some advice for the startup crowd.
He gave me something better: a free business idea, because he is just a great guy.
“I can’t stop thinking about the idea I heard, that someone could do the same thing as Ulpian, but for appliance manuals and user guides,” he said.
Imagine this: You ask, “How do I change the headlights on a Toyota RAV4?”
Instead of digging through a 400-page manual, AI gives you the exact answer — logically summarized, maybe with an image, maybe even with a custom-generated video.
Need to clean all your air filters? AI can list the steps in order. With parts. And safety tips.
Mic drop. Just brilliant!
If anyone runs with this, just make sure to send Filip a thank-you card, or send us a word, and we might be the first to invest.
Want to Use Ulpian?
If you’re a lawyer, legal researcher, or even just an AI nerd curious about how this stuff works in real life — check out Ulpian.com.
They’re currently expanding into new markets, onboarding beta users, and growing fast.
So if you want to reduce days of legal work into seconds — and actually make use of AI in a practical, high-stakes domain — Ulpian might be your new best friend.
And hey, maybe someday it’ll even make lawyer consultations cheaper too. 
Founding team:
Filip Boltužić
– PhD in NLP/AI and machine learning engineer (ex-Amazon AWS), co-created the core AI search logic.
Matej Gjurković – PhD in NLP/AI, with 16+ years in innovation and scaling systems.
Sandra Ćatović Stručić – Ex-Deloitte. Specialist in ESG, commercial law, and labor law. Over 5 years of experience working in a law office supporting general law practice.
Ozren Kobsa – Lawyer specializing in banking & finance, 15+ years in top-tier firms, STEM + MBA candidate
