What does it take to build legal AI inside a Big Law firm with a deliberately small team?

In Episode 2 of the LegalQuants podcast, host Victor Wang talks with Andrew Simon, Chief AI & Innovation Officer at Weil, Gotshal & Manges. Andrew explains how he evaluates new legal AI products, why he thinks retrieval is one of the hardest problems in legal AI, and why Weil uses both closed frontier models and open-weight models. He describes how lawyers’ prototypes become durable applications, why the firm chose decentralized innovation, and how BenchMark grew out of a client’s question.

The conversation then turns to Andrew’s path from Big Law to a startup, supply-chain mathematics, machine learning at BCG, and Weil, before closing on machine-to-machine legal work, access to legal knowledge, raising children in the age of AI, and what he won’t hand over to it.

Andrew Simon at Weil: https://www.weil.com/people/andrew-simon

LegalQuants Residency: https://residency.legalquants.com/

LegalQuants: https://www.legalquants.com/

Victor Wang: https://vic.law/

In this episode:

00:00 Welcome to the LegalQuants podcast

00:09 Meet Andrew Simon

01:12 How Andrew sizes up a new legal AI product

05:26 ROI when the technology keeps changing

06:37 Why legal retrieval is hard

10:40 Using closed and open-weight models

12:10 What happens when a new model drops

16:40 From a lawyer’s prototype to a real app

21:37 How Weil sees and supports AI use

24:14 Why Weil decentralizes innovation

27:16 Building BenchMark

31:31 How BenchMark caught on

33:07 The LASER name and the three-year goal

37:08 A startup within a Big Law firm

41:04 What Andrew looks for in a hire

44:57 Law school, Big Law, and leaving practice

49:21 Starting a beverage company

51:35 How supply-chain math led to machine learning

55:06 How Andrew came to Weil

58:38 The biggest lesson from BCG

59:26 Legal meaning, retrieval, and confidence

1:02:23 How strong a claim should a product make?

1:04:23 Agent security and machine-to-machine work

1:05:53 The case for an industry standards body

1:08:06 Where machine-to-machine efficiency matters most

1:10:27 Nonlawyers, AI, and knowledge diffusion

1:13:33 Raising kids in the age of AI

1:16:36 Parents, mentors, and curiosity

1:18:13 What he won’t surrender to AI

What happens when OpenAI asks practicing lawyers to help build a canonical set of legal plugins for the OpenAI for Law launch?

In Episode 1 of the LegalQuants Podcast, Victor Wang joins Jamie Tso and John Hughes to unpack how the team turned legal expertise into working plugins in just a few weeks. They share the ambitious first idea, the feedback that changed their approach, and what it takes to build something lawyers will actually use.

In this conversation, we discuss:

1. How a small WhatsApp group brought together lawyers who had been building alone—and became LegalQuants.

2. Why OpenAI pushed the team beyond an all-purpose “skill creator” toward concrete, customizable legal workflows.

3. How marketplace discovery changed the product: organize around litigators, transactional lawyers, and lawyers learning AI.

4. What makes a flagship skill: a real practice problem, a clear “aha” moment, testing with real documents, and a way to work with other skills.

5. How cite-checking, diligence, and redlining fit into larger acquisition, litigation, and negotiation workflows.

6. Why a legal AI skill needs a clear purpose even after the next model gets better.

7. Where to start: try a skill on work you understand, use legal design to make complex material easier to navigate, and feed what you learn back into the project.

8. Why Jamie sees legal quants becoming an internal capability at law firms—and what the Residency is building toward.

Try the LegalQuants plugins:

https://www.legalquants.com/lqplugins

Where to find LegalQuants:

• Website: https://www.legalquants.com/

• Residency: https://residency.legalquants.com/

• LinkedIn: https://www.linkedin.com/company/legalquants/

• Instagram: https://www.instagram.com/legal.quants/

• YouTube: https://www.youtube.com/channel/UCZen0gRIPZZlza8NMIg6otg

• Newsletter: https://legalquants.substack.com/

• GitHub: https://github.com/LegalQuants

Where to find Victor Wang:

• Website: https://vic.law/

• YouTube: https://www.youtube.com/@theviclaw

• LinkedIn: https://www.linkedin.com/in/stokebuilder/

• X: https://x.com/stokebuilder

• GitHub: https://github.com/stokebuilder

In this episode, we cover:

00:00 The OpenAI call

00:23 Welcome to the LegalQuants podcast

00:43 Meet the builders

00:59 How LegalQuants began

03:21 Community milestones and the Residency

04:42 OpenAI's legal plugin invitation

05:29 An ambitious first design

06:48 LQ Architect and feedback from OpenAI

07:38 John's cite-check skill

08:56 From specialized tools to broader practice

09:27 Who would discover the plugins?

10:52 Organizing the build

11:58 What makes a flagship skill

13:20 Three plugin families

14:01 How the skills connect

14:33 Acquisition workflow

15:17 Litigation workflow

15:58 Negotiation workflow

16:32 Open source and evolving skills

18:01 Where LegalQuants goes next

21:12 Practical advice for lawyer-builders

23:19 Thanks and where to follow

Referenced:

• LegalQuants Litigation, Transactional, and Companion plugins: https://www.legalquants.com/lqplugins

• LQ Architect: https://github.com/LegalQuants/LQ-Architect

#LegalQuants #LegalTech #OpenAI #Codex