Built by people who have spent 20 years in legal and 15 years in tech.

Referent is made by AI Lawtech - the team behind AI Lawyer, one of the most widely used consumer legal AI products. We build AI agents that run a law firm's routine, while the lawyer approves every critical step.

Why Referent exists

Referent didn't start as an idea on a whiteboard. It started inside AI Lawyer, our legal AI platform, where we watched lawyers use AI for exactly one thing: answers.

Answers are useful. But answers don't run a firm. The same solo attorney who used AI to draft in minutes still spent evenings on intake, deadlines, client follow-ups, and billing prep. Software stored that work. Nobody did it.

So we built the missing layer. Referent is AI-native legal practice management: AI agents handle intake, matters, client email, deadlines, follow-ups, and billing prep from your live matter context - and stage every critical action for your approval.

We are engineers and product people who have spent four years in legal tech, working alongside practicing attorneys. We know where AI helps a law practice. We know exactly where it must stop and wait for the lawyer. That line is the product.

What we believe

Leverage, not replacement

AI will not replace lawyers. It will multiply them. One attorney with agents handles the caseload of a small team - and the judgment stays human. We don’t sell replacement. We refuse to.

The lawyer stays in control

Anything client-facing or high-risk waits for your approval. Every agent action is logged in a full audit trail. AI prepares. The lawyer approves.

Confidentiality is a design constraint

Per-firm encrypted workspaces. No training on client data. Delete means delete. We built Referent in the EU, under the strictest privacy regime in the world, and it shows in the architecture.

The team

Senior engineers, product operators, and practicing attorneys. Based in Poland, building for US firms.

The engineering core averages 15+ years in technology: C-level and senior leadership roles at billion-dollar companies including Viber and Playtika, multiple CTO and founder track records with a successful exit, and systems shipped for Fortune 500 companies - including the lead engineer behind the AI Lawyer platform. This is the team's second legal AI product, not its first.

Referent is led by founder and CEO George Zaletski - 15 years in tech, the last four in legal tech, with earlier product and growth work for brands including PayPal, Nestle, and Kaspersky.

We have been building AI legal tech since 2022 - before legal AI was a category with a name. Four years of shipping AI to lawyers taught us what most of the market is still discovering: where AI genuinely helps a practice, and where it must stop and wait for the lawyer.

George Zaletski, Founder and CEO of Referent

Our vision

Lawyers bring judgment, strategy, and care. AI brings speed, scale, and consistency. Referent exists to put the two together - so a solo or small firm can carry far more matters without losing the craft that makes the work worth paying for.

I spent fifteen years building technology, time inside a law office, and four years in legal tech. The pattern never changed: capable lawyers buried in operational work that software should have handled. Referent is the system I wished existed - it runs the routine and hands every client-facing decision back to the lawyer.

George Zaletski

Founder & CEO, Referent

Recognized by

Backed by Google for Startups and NVIDIA Inception.

Featured in Yahoo Finance, AP News, Business Insider, MarketWatch, MSN, and Digital Journal.

Company

Product
Referent - AI-native legal practice management software for US solo and small law firms.
Company
AI Lawtech Sp. z o.o., registered in Poland (EU).
In legal tech since
2022. Also the team behind AI Lawyer.
Infrastructure
Google Cloud Platform, with US and EU data residency. See our Security page

Join the next
generation of AI-native lawyers

Private beta for selected lawyers and law firms.

Apply for Private Beta

Personal invite

White-glove onboarding

No data training