Legal AI Lab

What does an AI-native law firm look like if you build it from scratch today? Arvid van Oorschot and Jelle Drok left their partnerships to find out.

Both spent more than twenty years litigating patent and technology cases at the top of the Dutch market. Jelle was a partner at Freshfields, Arvid at the IP boutique Vondst. In September 2026 they opened Driven Lawyers in Amsterdam, which they call the first AI-native law firm in the Netherlands.

They built their own AI drafting software and work with fixed fees instead of the billable hour. The lawyer stays in the driver's seat. They do the thinking first, then AI writes the first draft. Their experience sits in about a hundred frameworks, built the way a senior lawyer instructs a junior.

Arvid and Jelle explain how you put a fixed fee on complex litigation, why they let AI attack their own arguments, and who trains the juniors in an AI-native law firm. Jelle also makes a bold prediction about the rest of the market.

You'll learn

• why drafting was the first thing they took out of the old law firm model
• how you set a fixed fee for litigation when the other side can file a counterclaim
• what the under-the-shower moment is, and why they bring it forward
• how a hundred frameworks turn twenty years of experience into instructions for AI
• why one bad argument can ruin a whole case
• how juniors learn in a firm built on the old guild model of master and apprentice

Chapters

(00:00) Introduction

(02:35) Clients want to kill the billable hour

(05:54) Building an AI-native law firm from scratch

(07:46) Creation, production and the shower moment

(11:09) A fixed fee for complex litigation

(17:12) An AI-native law firm with a human motto

(18:49) Talking to investors about a scalable model

(22:10) Twenty years of experience in frameworks

(26:26) Vibe coding the MVP with Claude

(30:42) Zeno's question: the hardest part to put in a framework

(34:50) How juniors learn in an AI-native law firm

(40:39) Better work at a lower price

(43:50) The devil's advocate tool

(45:42) Leaving the partnership

(48:39) Why every law firm will work like this

(50:43) Advice for managing partners

(52:29) Own AI models and your own knowledge

(55:50) Advice for young lawyers

Powered by Zeno

This episode is powered by Zeno. Zeno's AI navigates law like a human legal professional, grounded in authoritative sources, backed by traceable citations. Visit zeno.law and make deep thinking your competitive edge.

Legal AI Lab is a podcast by Recht in je Oor.

Want to advertise in this podcast? Visit rechtinjeoor.nl/adverteren for all options.

Hosted on Acast. See acast.com/privacy for more information.

AI breaks the law firm business model that partner compensation was built on.

Jaap Bosman is founder and CEO of TGO Consulting and author of Law Firm Partner Compensation, written with Jaime Fernández Madero. He advises law firms across Europe, the Americas and Asia.

His starting point is the creation production divide. Clients pay for the moment a partner sees the solution. Firms bill for the hours of production that follow, and those hours quietly pay for the creation. So who pays for the creation when machines shrink the production?

Bosman explains why AI is the great equaliser once every firm uses Harvey, and why only a handful of firms build their own AI, like Kirkland & Ellis with Palantir and Latham & Watkins with NVIDIA. He shares his research into what clients praise in their best lawyers, and what partner compensation says about the ambition of a firm. Later in the conversation he explains how his clients move away from the billable hour step by step, without panic.

You'll learn

• why partner compensation is about culture and ambition in the end
• how production hours have paid for creation, and what happens when AI shrinks them
• why AI is the great equaliser once every firm uses the same tools
• what the seven core dimensions are that set top ranked lawyers apart
• why fixed fees only pay off when partners learn to budget time
• how a firm leaves the billable hour without a six month panic

Chapters

(00:00) Introduction

(02:59) From C&A to NautaDutilh: an unusual career path

(11:41) Partner compensation defines what a law firm is

(13:17) The champion and the twin sister in the audience

(16:09) Why the law firm business model is starting to unravel

(19:46) The creation production divide

(24:07) The Picasso story: who pays for creation?

(26:02) Pricing, the value matrix and the hourly assumption

(29:57) Kirkland, Latham and proprietary AI

(31:17) 83% of client praise is about human skills

(36:23) The seven core dimensions of top lawyers

(45:35) A law firm business model built on human skills

(47:31) Freshfields and the AI bifurcation in Big Law

(50:18) Zeno's question: AI for smaller law firms

(54:22) Delegation and the rubbish question problem

(56:45) The billable hour in the age of AI

(1:02:07) Value-based billing and plain vanilla work

(1:06:34) A historic opportunity for mid-tier firms

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law. Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources. Visit zeno.law and make deep thinking your competitive edge.

Legal AI Lab is a podcast by Recht in je Oor.

Want to advertise in this podcast? Visit rechtinjeoor.nl/adverteren for all options.

Hosted on Acast. See acast.com/privacy for more information.

Most law firms buy their legal AI. Freshfields builds its own, in a lab in Berlin.

In this Legal AI Lab Special we went to Berlin for a tour of the Freshfields Lab. Lawyers, developers and data scientists work side by side there to build the AI tools the firm uses worldwide. It started with a hard problem. During the German diesel mass claims, paper arrived in every office and every conference room was filled with people typing it into Excel sheets. Today that runs through one secure platform that connects ChatGPT, Claude and Gemini, and adds the firm’s own knowledge, templates and skills on top.

Freshfields is the only law firm working directly with Anthropic, after starting with Google Cloud. That gives the lab access to engineers and to models that are not yet released. In return, the AI companies learn what a law firm needs: very long context, heavy document uploads and data separated per client.

The team also explains why they built instead of buying Harvey or Legora, why leaderboards for skills and usage work, and why they see the real value in better work rather than faster work. Juniors use AI as a tutor, and the people who build the tools see lawyers lining up with ideas.

You’ll learn

• why one platform with every major model beats committing to a single vendor
• what makes a general LLM useful for lawyers: firm knowledge, templates and skills
• what Freshfields gets from working with Anthropic and Google, and what they learn in return
• why AI changes the scope of a due diligence, not only its speed
• how juniors use AI as a sparring partner instead of calling a colleague

Chapters

00:00 Welcome to the Freshfields Lab in Berlin

00:51 From paper rooms to AI: the diesel mass claims

01:52 One secure platform for every model

02:31 Why the lawyer’s knowledge makes the difference

03:00 Leaderboards for skills and usage

03:19 Partnering with Google and Anthropic

04:40 What a law firm asks of an AI lab

05:10 Building instead of buying Harvey or Legora

06:08 Better work, not just faster work

06:51 What lawyers get out of it

07:26 AI as a tutor for juniors

08:12 More technologists inside law firms

08:40 Universities that still ban AI

Hosted on Acast. See acast.com/privacy for more information.

Freshfields did not buy a legal AI tool. It built a toolkit around all of them.

Saloua Ouchan is a partner at Freshfields in Amsterdam and co-head of the firm’s AI efforts. Freshfields has had its own lab since 2019, where lawyers, data scientists and machine learning experts build tools together, and it develops them with both Anthropic and Google. Every lawyer gets their own AI toolkit with a dropdown of the major models and a set of skills for recurring tasks. The firm will never build its own LLM. What it keeps for itself is two centuries of accumulated knowledge from cross-border deals.

That shifts the pressure to the junior lawyers. The data crunching that used to train them is disappearing, and polished AI output makes it easy to hand in work nobody has thought through. Ouchan explains why the apprentice model matters more now, and what she does when a junior can’t explain their own due diligence report.

She also covers what clients ask about confidentiality, why the billable hour is not dead yet, how internal leaderboards changed the culture, and why she thinks the Jevons paradox means more work for lawyers, not less.

You’ll learn

• Why Freshfields chose a vendor agnostic toolkit over a single legal AI subscription
• How skills replace prompting for tasks you perform ten times a week
• Why juniors are hit harder by AI than partners, and how the apprentice model adapts
• Why critical thinking matters more when the output looks this convincing
• How due diligence can be priced in three tiers, from fully automated to a full lawyer check
• What tracking AI use per client matter means for future billing models
• Why AI could unlock latent demand for legal services

Chapters

00:00 Introduction

00:55 Freshfields’ own AI toolkit

01:40 A vendor agnostic choice of models

03:13 From prompting to skills

04:16 How AI changes the work of a partner

05:10 Are juniors losing their training ground?

07:03 The apprentice model when the groundwork is automated

09:24 Don’t be fooled by well written AI output

10:29 Building in house instead of buying a tool

12:35 What clients ask about AI and confidentiality

15:06 Is the billable hour disappearing?

16:04 Three flavours of AI due diligence

17:28 Pricing a product instead of writing down time

19:04 Co-creating with Anthropic and Google

21:09 New roles inside the law firm

21:43 Why Freshfields did not rush to the press

23:30 Lawyers and technologists in one room

24:07 Clients inside the lab

24:52 Leaderboards and making AI fun

27:37 Will young lawyers still have a job?

Hosted on Acast. See acast.com/privacy for more information.

Microsoft is not trying to win the legal AI market. It wants to be the layer underneath it.

Dervish Tayyip is Assistant General Counsel for EMEA at Microsoft. The strategy he describes has little to do with building the best model. Embed intelligence into the workflows lawyers already use, then wrap enterprise governance around it. The legal agent inside Word is the visible part. The orchestration layer, where a lawyer pulls down whichever model suits the task, is where Tayyip says legal AI value actually sits.

For general counsel the pressure is coming from somewhere else entirely. CFOs have heard that legal is about to be disrupted. Management consultants are telling the CEO they can deliver the same legal department at a third of the cost. Tayyip explains what he hears in Microsoft's roundtables with general counsel, and how a GC takes control of that narrative before someone else writes it for them.

He also covers where horizontal platforms end and jurisdiction specific legal AI begins, what Microsoft commits to if a government ever demands a shutdown, and what junior lawyers should be learning now the groundwork is automated.

You'll learn

• why Microsoft says the frontier model is not where legal AI value sits
• how the legal agent inside Word removes the friction of changing behaviour
• what enterprise governance adds on top of a model: confidentiality, auditability, privilege
• how a general counsel builds the investment case when the CFO wants headcount down
• why horizontal platforms and vertical legal AI are no longer an either or choice
• what Microsoft commits to contractually if a government demands a shutdown

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law.

Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources.

Visit zeno.law and make deep thinking your competitive edge.

Chapters

00:00 Introduction

02:18 Why the legal profession adapts slower than technology

10:39 Automate, innovate or eliminate your workflows

14:14 The legal agent in Word and an operating system for legal work

17:11 Why the model is not where legal AI value sits

22:39 Competing and partnering in the legal AI market

25:56 Inside Microsoft's roundtables with general counsel

29:59 What if someone pulls the plug on Europe

33:48 200 data centres and Microsoft's promise to Europe

43:10 The general counsel's case to the CFO

46:30 Where horizontal and vertical legal AI meet

52:43 How junior lawyers learn when the groundwork is gone

58:23 Efficiency versus effectiveness

Hosted on Acast. See acast.com/privacy for more information.

AI is not neutral. The gender gap in AI runs through the data it is trained on, the people who build it and the people who use it. And in legal, almost nobody talks about it.

Nicky Leijtens, co-founder of She Breaks the Law and part of Bird & Bird, joins Hidde Bruinsma with examples that stick. Ask ChatGPT for career advice as a girl and the answer is less encouraging than the one a boy gets. Women make up 22% of AI development teams, a number that has not moved in ten years of Stanford research. At the M&A Awards, 42 nominations included exactly one woman.

Why does the gender gap in AI stay off the agenda while every conference talks adoption and implementation? And what can you do tomorrow, whether you build, buy or simply use legal AI?

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law.

Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources.

Visit zeno.law and make deep thinking your competitive edge.

You'll learn

• why AI reflects the people who build it, and who is missing from that room
• what happens when biased models start influencing legal decisions and access to justice
• how visibility creates permission for other women to experiment
• why women adopt AI less and are judged more harshly when they do use it
• what closes the gender gap in AI according to Nicky: small actions that add up

Chapters

00:00 Introduction

00:05 She Breaks the Law: 4,000 women in 50 countries

07:43 Rechtwijzer and better justice journeys

13:51 The gender gap in AI nobody talks about

14:39 AI is not neutral: data, users, builders

16:39 Airbags, heart attacks and the male default

18:08 Only 22% of AI builders are women

20:25 Closing the gender gap in AI: where to start

25:30 The Rik de Groot moment at Legal Geek

29:14 Zeno question: auditing legal AI tools for bias

34:27 42 nominations, one woman at the M&A Awards

42:54 Antibodies: who is blocking the change

50:39 What you can do tomorrow about the gender gap in AI

Hosted on Acast. See acast.com/privacy for more information.

Personal branding is career insurance for lawyers. That is how George Hannah treats it, and it is paying off.

George is a solicitor apprentice at Lewis Silkin and the founder of Best Practice, the legal AI newsletter he started in January 2025 with one viral LinkedIn post. Today it is a media company with a podcast and a personal brand of more than 12,000 LinkedIn followers, run entirely next to his day job.

We get into his personal branding playbook: why perfectionism keeps lawyers invisible, why an email list beats a rented audience, and why he posts at 80 percent and walks away. Then the legal AI market: Jude Law fronting Legora, Gabriel Macht fronting Harvey, Claude for Legal, the Freshfields deal and the Gartner hype cycle.

And the question every firm is wrestling with: what is left for junior lawyers when AI does the grunt work? George's answer: build the agents.

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law.

Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources.

Visit zeno.law and make deep thinking your competitive edge.

You'll learn

• why a personal brand works as career insurance
• how one viral post grew into a legal media company
• why perfectionism keeps lawyers invisible on LinkedIn
• what the celebrity brand war says about the legal AI market
• whether legal AI has reached the peak of the hype cycle
• how junior lawyers stay valuable by building AI agents

Chapters

00:00 Introduction

02:21 One viral post starts Best Practice

04:49 Email lists and the discipline of weekly content

08:09 From Sunday newsletter to legal media company

11:19 Personal branding is not luck

12:22 Why lawyers fail at personal branding

16:03 Personal branding in the age of AI slop

18:07 Your personal brand as career insurance

22:01 The TBD Marketing LinkedIn leaderboard

23:39 Post at 80 percent and put your phone away

26:06 Jude Law, Gabriel Macht and the legal AI brand war

28:58 Have we reached the peak of the hype cycle

30:30 Claude for Legal, Microsoft and the Freshfields deal

35:47 Access to justice and Lawhive's AI lawyer Lawrence

40:32 What junior lawyers do when AI takes the grunt work

47:08 Portfolio careers and weekly check-ins

50:50 Advice for lawyers finding their voice

Hosted on Acast. See acast.com/privacy for more information.

Law firm disruption won't start with technology. It starts with the client. Bjarne Tellmann spent 25 years as general counsel at Coca-Cola, Pearson, and GSK. His book "Law in the Era of AI" maps how the changes inside corporations will force the legal industry to adapt or vanish.

Firms are posting record profits. But 100% of GCs in a recent Axiom survey regret their law firm engagements. 89% no longer consider them adequate. Bjarne draws on Nokia, Kodak, and Clayton Christensen's disruption theory to show why the most profitable incumbents are the most exposed to law firm disruption.

He also offers paths forward. ClearyX funded its own disruption. The gazelle elephant model shows GCs how to rebuild their teams. And a warning: agentic AI is already making autonomous decisions inside corporations. If general counsels don't step into governance now, they'll be cleaning up the mess later.

You'll learn

• why law firm disruption always starts with the client, not technology
• what the milkshake analogy teaches about law firm blind spots
• why no law firm has ever gone under while still profitable
• how ClearyX funded its own disruption on purpose
• what the gazelle elephant model means for legal team design

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law.

Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources.

Visit zeno.law and make deep thinking your competitive edge.

00:00 Introduction

02:15 Why innovation needs ideas from unexpected places

08:50 Record profits but law firm disruption is coming

12:20 The AI factory driving law firm disruption

17:10 The profession with zero curiosity about its clients

23:00 Jobs to be done: what clients actually hire lawyers for

32:00 Why law firm disruption is structurally impossible

38:20 What if tech companies answer legal questions

42:15 Kodak invented digital photography and still failed

46:30 ClearyX: a firm that funded its own disruption

53:10 The gazelle elephant model for legal teams

57:30 Agentic AI and why governance cannot wait

Hosted on Acast. See acast.com/privacy for more information.

A banking lawyer at one of the world's largest law firms teaches herself Python on weekends. She builds an AI-powered billing tool in a Jupyter notebook. Months later, that tool is rolled out across the entire firm.

Tanya Sadoughi is Linklaters' AI and Innovation Lead Lawyer for Global Banking. When ChatGPT launched one month into her innovation secondment, she saw the opportunity before most others did. She talked to every lawyer in the banking practice, mapped every pain point, and built a working prototype herself.

That prototype became the WIP Summariser, now deployed firmwide. Today she leads a team of five AI lawyers inside a global network of twenty.

Vibe coding is changing how lawyers work. Not by replacing them, but by turning them into builders. Tanya explains what that looks like inside a major international law firm. From hackathons that spark curiosity to agentic systems that solve problems at scale.

You'll learn

• Why the AI lawyer role should always be filled internally
• How a weekend prototype became an enterprise tool at Linklaters
• What vibecoding means for the legal profession
• Why hackathons are the best way to spark AI curiosity at law firms
• How Linklaters scaled from one to twenty AI lawyers
• What lawyers and engineers can learn from each other

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law.

Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources.

Visit zeno.law and make deep thinking your competitive edge.

0:00 Introduction

2:04 Will lawyers become builders

4:53 Growing up in a pizza shop in South Yorkshire

9:17 A viral YouTube channel at 15

11:16 Why she chose law

13:29 Becoming a banking lawyer at Linklaters

15:42 The itch for something different

17:35 ChatGPT launches one month into her innovation role

19:28 First experiments with large language models

23:18 The billing problem nobody could solve

27:26 Learning Python on weekends

31:31 The moment the prototype worked

35:31 Hackathons and sparking curiosity at law firms

40:23 From one AI lawyer to twenty

43:47 Vibecoding five apps in one week

46:31 What engineers and lawyers can learn from each other

51:23 Why every law firm needs an AI lawyer

53:01 More builders not more tools

Hosted on Acast. See acast.com/privacy for more information.

Lawyers are architects of society. But what happens when AI changes how law is accessed, practiced and understood?

In this episode of Legal AI Lab, Soledad Atienza, Dean of IE Law School, explains why legal education must move beyond national systems and embrace a global legal mindset.

AI makes legal knowledge more accessible than ever. That means law schools must shift their focus. From memorising content to developing critical thinking. From national silos to comparative principles. From narrow specialisation to multidisciplinary awareness.

We discuss:

• Why law schools should not ban AI but guide its use

• How assessments must change in an AI world

• Why students must learn to question AI outputs

• Why demand for legal services is growing, not shrinking

• The importance of human skills such as empathy, negotiation and teamwork

Powered by Zeno

This episode is powered by Zeno. Zeno is an AI native legal workspace built for Dutch and EU law.

Its AI navigates law like a human legal professional. Secure, transparent and grounded in authoritative sources.

Visit zeno.law and make deep thinking your competitive edge.

00:00 Lawyers as architects of society

02:47 National regulation vs global practice

06:18 From systems to principles

09:42 Knowledge vs judgment in the AI era

14:11 Growing demand for legal services

17:36 Embedding AI across the curriculum

22:04 Teaching students to question AI

26:53 Rethinking assessment in an AI world

32:27 Generalists and multidisciplinary thinking

37:12 Human skills in a tech driven profession

41:05 Experiential learning and VR

44:38 Habits future lawyers must build

Hosted on Acast. See acast.com/privacy for more information.