--- title: "The GC as Fashion Designer: Tailoring Legal Advice in the AI Era with Addar Levi, Lazard Deputy GC" description: "Addar Levi is Global Head of Regulatory Affairs and Deputy General Counsel at Lazard." canonical: "https://gc.ai/podcast/the-gc-as-fashion-designer-tailoring-legal-advice-in-the-ai-era-with-addar-levi-lazard-deputy-gc" last_updated: "2026-11-10" --- # The GC as Fashion Designer: Tailoring Legal Advice in the AI Era with Addar Levi, Lazard Deputy GC > Addar Levi is Global Head of Regulatory Affairs and Deputy General Counsel at Lazard. ## What does it mean to be the GC as fashion designer? Addar frames outside counsel advice as high-quality raw fabric that the in-house lawyer must reshape to fit the actual business situation. The memo from a white-shoe firm is carefully sourced material, but it arrives cut for a generic wearer, and the GC's job is to tighten it, loosen it, or redrape it entirely until it fits. The clearest illustration she offers is a diligence memo that flagged a litigation risk on page 12 and noted the company would ultimately prevail. With no one to lift that risk out of the prose and present it in terms the board actually cared about, the board moved forward without understanding what 'ultimately prevail' would cost them along the way: > The danger of not having the tailor, the fashion designer GC, is that the board just glossed over the risk on page 12. But of course, that risk ended up materializing. And it turned out that even if we were ultimately going to win, we were going to lose on a motion to dismiss. We were going to lose a preliminary injunction. It was going to create huge delays and huge costs. ## What should a one-slide board memo on litigation risk look like? The Prada version of a 12-page diligence memo is a single slide that names the three risks that matter most and plays them out concretely. That means translating 'you will ultimately prevail' into what the litigation actually costs in time, money, and operational disruption before the favorable outcome arrives. The purpose of that translation is to ward off the question that always follows a bad surprise: > The question always inevitably is, well, why didn't the lawyers tell us? It's trying to ward off that. ## How should a new in-house lawyer approach the transition from a law firm? Addar identifies two adjustments that matter most when moving from outside to in-house work. First, the shift from being a lawyer first to being a business or policy person first: understanding the substance of what the client is actually trying to accomplish has to precede the legal analysis. Second, listening well enough to understand what question is actually being asked. She describes herself as a lifelong closeted introvert and suggests that orientation makes it easier to prioritize input over output: > Your output is only as good as your input. And so I learned the importance of really listening in an in-house role to really understand what it is your client's telling you. ## How do you identify the question beneath the question? The stated question from a CEO or board is rarely the complete question. 'Can we do this?' might mean how to do it, what could go wrong, or simply a request for validation. Addar says the answer requires reading both the words and what the room is communicating nonverbally: > Sometimes they're actually asking you how do we do this? Sometimes they're asking you, what's gonna go wrong here? Sometimes they're just asking you, can you tell me I'm right? And I think really listening both to what they're saying, but also the body language, and I think this applies equally with your team, is the key to actually answering the question beneath the question. ## Why does AI-generated legal work product create a new risk of its own? The specific danger Addar identifies is that a well-formatted AI draft looks finished when the substantive legal work has not happened yet. Before AI, the blank screen signaled that the hard thinking was still ahead. Now, a draft with headers and coherent structure arrives immediately, and that appearance of completeness can cause people to treat it as an endpoint. She connects this directly to advice from her law professor Jack Goldsmith, who told his students to read the statute rather than rely on summaries or other lawyers' analyses of it. The same principle applies to AI drafts: the subject-matter knowledge required to evaluate and refine the output cannot be shortcut: > Pre-AI you were further along pretty far along once you had a fully formed first draft. Now you get this first draft and it's appealing right? It has the headers, it looks really compelling, and the thing I try to tell myself and my team is that's when the work starts now. ## What did leading 2,000 attorneys at the Treasury teach you about scale? Moving from a team of about 10 at the Obama family office to acting GC of the Treasury was, in Addar's description, a total crash course in scaling leadership. She notes that the fundamentals of management do not change, but a small team allows a leader to cover for weak habits in ways a large one does not. At scale, the job becomes setting clear strategic goals, facilitating information flow, empowering people to do good work, and exercising careful judgment about which issues require personal attention: > You can sort of fake your way through a 10K, but you can't fake your way through a marathon. So I think when it's a smaller team, you can still kind of be a micromanager. But when you have a team that's 2,000 attorneys, you can't be a micromanager. ## What does government service teach lawyers about operating under pressure? Addar describes a structural difference between government legal work and private-sector legal work: the government often cannot choose to do nothing. That constraint forces a posture of selecting among imperfect options and communicating them clearly, which builds both analytical calm and messaging skill. The Silicon Valley Bank weekend in March 2023 illustrated how compressed that decision cycle can be. From the initial deposit run on Thursday to a completed acquisition by the time Asian markets opened Monday, every option had to be evaluated and acted on in real time: > The US government doesn't always have the privilege of doing nothing. Right. You often actually have to be in the stream of commerce, as it were, every single day. And so you're in this sort of posture of having to pick from least worst options. Which I think helps you hone sort of analytical skills and sort of calm under pressure, as well as messaging, right? How do you then message that least worst option?