What this covers

This conversation is between Shane Parrish and Winston Weinberg, co-founder and CEO of Harvey, the AI platform for the legal industry. (The transcript's speaker labels are garbled, but the statements are attributed to Weinberg.) It runs in two halves. The first covers how Weinberg runs the company: prioritization, decision speed, stress, and hiring. The second covers what AI does to legal work and to the people who do it. Harvey's origin story sits between them. Weinberg describes testing GPT-3 on real legal questions, and a cold email to Sam Altman at OpenAI. He also describes a demo strategy of building each pitch around a prospective lawyer's own recent work product, which made the capabilities tangible.

The operating advice is specific. Weinberg treats nearly every decision as a two-way door, reserving the one-way door label for the rare irreversible case. He argues people struggle to say no because yes brings immediate approval while the payoff from no arrives months later. He says leaders should rebuild their priority list every three to six months. His view of resilience is that a team that has failed and recovered together is worth more than any external threat level. On urgency, a founder cannot personally sustain it across 800 people, so it has to come from hiring people who already feel it. He also describes stress management through sleep and morning runs.

The legal material is where the claims turn contentious. Weinberg splits professional work into "work done," which gets automated, and advice and judgment, which gain value. He expects lockstep associate promotion to give way to skill-based advancement, and he thinks law firms may not shrink but will need fewer people per project. He explains why all-AI firms are blocked by unauthorized-practice rules and ABA restrictions on outside ownership, outside the Arizona and Utah sandboxes. He also predicts AI regulation will create new legal work, and he argues for redesigning the second and third years of law school around hands-on experience.

Sharpest takeaway

Winston Strawn argues that startup founders must ruthlessly prioritize the company's core bottleneck, build resilient teams through stress-tested decision-making, and recognize that AI amplifies skill differences, requiring lawyers and professionals to shift from commodity work to high-judgment client advisory.

  • Founders succeed by identifying and obsessing over the single most important problem while ignoring everything performing well
  • Resilience comes from normalizing failure early and often, not from avoiding stress
  • AI will commoditize execution work but amplify the value of judgment, relationship-building, and client understanding

The claims · ranked64 claims · weighted by value

This asset isn't compiled yet

You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.

0.69

Sleep and morning exercise are the primary mechanisms for maintaining personal stress management and CEO performance; morning runs provide a daily reset that improves emotional regulation for the rest of the day.

normativehigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“I like getting enough sleep and working out are the main things for me. It's it's not even close. If you wake up right in the morning and you go on a run, I found that that's like incredible because everything else through the rest of the day you get like a a shock, basically, and then the rest of the day you're like calm her for everything else.”

0.69

Urgency scales through hiring and culture; a founder cannot personally maintain urgency across 800 people, so the only scalable approach is to hire people who already feel urgency and train them to instill it in their teams.

causalhigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“The best way to do this is hiring. There's no other do it. Um it is hiring and culture. And so, what you need to do is hire a very good team that all feels the same way you do. And what your job then is is how do you instill a sense of urgency into the rest of that team? And then those leaders will instill it into their team, and then those leaders will instill it into their team, right?”

0.69

Building a company requires making a thousand small bets, only a few of which succeed; people avoid this because they treat each bet as a failure rather than as a necessary part of the process.

causalhigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“building a company is a thousand failures and then a couple successes. And I think what people don't realize is that you have to fail constantly in order to succeed at bigger things, right? Like you have to make a million bets and then one of them pays off.”

0.69

Regulatory oversight of AI use will be inevitable, requiring lawyers to monitor and ensure compliance, creating new legal work rather than reducing it.

forecasthigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“We are going to have a bunch of regulations about AI use, for sure. There's no universe in which we aren't going to end up there, right? In the same way we have a lot of audit restrictions on banks, right? For like when you make a trade, you have to put XYZ down. That's going to happen for models, for sure, right? Who's going to be in charge of that, monitoring that? Lawyers, right?”

0.69

Unauthorized practice of law rules and ABA ethics rules against non-lawyer investment in law firms currently prevent all-AI law firms except in Arizona and Utah regulatory sandboxes.

factualhigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“you can't do that in the US right now. Um so, you need a lawyer. Basically, there there's two rules. Um one is state by state, which is the unauthorized practice of law. It's actually a felony. Um and so, you know, you actually need to be barred in order to give legal advice. Um and then the other one is an ethical rule. So, it's by the ABA, um which basically says that you can't have in uh non-lawyers invest in a law firm.”

0.69

Principle 2 of running a company: Right people in right positions is critical; this includes hiring, firing, but also mentoring, role fit, and team alignment.

normativehigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“Number two is do you have the right people in the right positions? Um like that is just so important... That's not by the way just like hiring and firing at all. That's also like is this person the right fit for this role? Are you doing a good job of mentoring this person?”

0.69

The next 1-2 years will define which companies are successful for the next decade or longer, making current period business decisions existentially important.

forecasthigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“the next, you know, year or two years are basically going to define the companies that are successful for the next decade, probably, right? If not more.”

0.69

Law firms have 'rainmaker partners' who capture disproportionate work due to slight skill/speed advantages; they are probably only 10-20% better at specific tasks, but this compounds into capturing most work.

factualhigh valueestablishednovelty 1/4durability 3/4· Winston Strategist

“You have this thing at a lot of the law firms where, you know, you call them rainmaker partners. And you know, there's an argument of like how many how much better are they at every single task than everyone else? Like probably not that much better, but they're slightly better at a couple things and then those things compound. And then the vast majority of work goes there.”

0.68

Leaders must completely redo their prioritization approach every 3-6 months because company needs fundamentally change, and failure to do this creates organizational dysfunction.

normativehigh valuecontestednovelty 2/4durability 3/4· Winston Strawn

“every, you know, like 3 to 6 months, you have to completely redo how you do prioritization as a leader, I think. Um and if you don't, you're really going to start messing things up.”

0.68

Professional services contain two types of work: (1) 'work done' (deliverables like contract review) which gets fully automated, and (2) 'decision/advice' which becomes more valuable as automation removes commodity work.

definitionhigh valuecontestednovelty 2/4durability 3/4· Winston Strawn

“There are there is I have work done, which is like I deliver a work product, right? And that is going to get fully commoditized... The decision or advice is not... In fact, my guess is it's actually going to be more valuable over time.”

0.65

Principle 1 of running a company: Founders should spend almost all time on product once distribution exists, because product is the only thing that scales; stepping away from product to drive sales is a mistake.

normativehigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“once you get to a certain level of distribution... founders should spend almost all of their time on product. All like, as much as possible. Um, the biggest mistakes I've ever made as a as an exec are, I step away from product and I try to make up for it with doing sales. Terrible, terrible idea.”

0.65

Decision-making principle: First determine if something is a one-way door or two-way door; 99.9% of decisions are two-way doors, meaning they are reversible.

normativehigh valueestablishednovelty 1/4durability 3/4· Winston Strawn

“determine if something's a one-way or two-way door immediately. Like that's the first thing, right? Um with the idea that baseline 99.9% of things is a two-way door.”

0.64

First year of law school teaches critical thinking, argumentation, and research skills by studying old cases, which is valuable regardless of specific law taught; second and third years need redesign to focus on hands-on experience.

normativehigh valuecontestednovelty 2/4durability 3/4· Winston Strawn

“The first year of law school is, I think, perfect... most of the cases you read about and the law that you read about is not good law anymore... But, why it's helpful is it teaches you how to critically think. It teaches you how to make arguments. It teaches you how to do research.”

0.64

Models like GPT-3 and early LLMs still hallucinate frequently, which caused early testers to dismiss them after a single bad output, creating selection bias where only people willing to tolerate failure continued using them.

factualhigh valueestablishednovelty 1/4durability 2/4· Winston Strawn

“Super risky because you know a decent amount of time it would hallucinate and get something wrong. And if it hallucinated the lawyer was like okay this is useless like get me off this demo. But when it worked it worked super well.”

0.63

Law firms currently promote associates based on seniority (lockstep) regardless of skill differences until year 8-9 partner decision, but AI will force them to change to skill-based, faster promotion for exceptional talent.

forecasthigh valuecontestednovelty 2/4durability 2/4· Winston Strawn

“right now in in law firms, it's basically lockstep. So if you're better than everyone else in your class, you get promoted at the same rate everyone else does. Until like year eight or nine and then they decide like if you're going to be partner or not, right? And I think law firms are going to they're going to have to change this.”

0.63

Law firms may not get smaller overall because AI allows institutionalizing knowledge faster and scaling across more projects, but will need fewer people per project, potentially enabling more projects.

forecasthigh valuecontestednovelty 2/4durability 2/4· Winston Strawn

“I don't know if law firms are necessarily going to get smaller because there's actually a world in which because AI allows you to institutionalize your knowledge faster and work across a a lot of different projects faster, you might actually be able to scale faster as a law firm than you used to be able to in the in the past, but you will need less people per project, right?”

0.61

People have difficulty saying no to requests because they receive immediate positive feedback (appearing responsive, collaborative) from saying yes, while the real gains from saying no are delayed 6+ months and invisible to external stakeholders.

causalhigh valuespeaker onlynovelty 2/4durability 4/4· Winston Strawn

“The reason I think people have a really hard time saying no is they say yes to a lot of things that make them look good in the short term, instead of being able to take the pain of ignoring it, and then doing so well because you ignored it, that people back off.”

0.52

Professional services are cyclical (legal work expands with M&A activity) but people analyze them in a vacuum; understanding what products companies will build in 10 years (50x more shipping capacity) changes the analysis of legal work.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“professional services are cyclical and counter-cyclical, but mostly cyclical, right?... they miss is it can do legal work today. What does legal work look like in 10 years? Like, what is the product review for a lawyer look like when Meta can ship 50 times as many products daily, right? That is going to change a lot of this world.”

0.52

AI technology is changing at a pace that professionals outside the tech industry are not mentally equipped to track, making it hard for them to build mental models for an environment where every product release is a massive step change.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“that rate of technology change people aren't really used to. And especially folks that are not in the tech industry, right? And with what's happening in AI right now, every release is a massive step change. And I think it's hard for people to wrap their mental models around that and think about, 'Oh, how do I create How do I become, you know, anti-fragile where I'm okay with making different choices and really think about the future when everything is changing so quickly?'”

0.52

Team resilience (having repeatedly failed together and recovered) is more valuable than external threat level for predicting company success, because resilient teams can adapt to any problem while fragile teams break when facing novel challenges.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“I now have a team of people that I have failed with so many times. Like we have taken crazy risk. We have massively butchered product lines and then recovered them and made them much better. We have made the wrong hires. We've tried to buy a company super early on that would have collapsed us... And because of that, I just feel very confident in my team's like resilience and their ability to adapt and strive and get through the next problem.”

0.52

The speaker makes decisions very quickly and has no regrets about speed; their only regrets involve not making decisions fast enough, missing windows for action.

factualhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“everything that I regret at the company or whenever I regret something, it's not making a decision fast enough. It's not I made the wrong decision. What I've always regretted is sitting at the bottom of the stairs for too long instead of jump up a stair and then I'll figure it out and then jump up a stair and I'll figure it out.”

0.52

As CEO, the speaker recognizes that their stress levels directly cascade through the organization; when they get overly stressed about something, the entire 800-person company responds, creating thrash and confusion about what actually matters.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“When you're an 800-person company if the CEO gets too stressed about something, the whole company will move like based off of that stress. And that is such a big problem. Like that's how you create thrash. You create thrash by you get overly stressed about everything and then people don't know... what actually matters, what doesn't matter.”

0.52

People often avoid step-by-step decision-making because planning all steps up front feels safer than taking steps and adapting based on new information, even though adaptive approaches succeed more often.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“It's less scary. So I think there's actually another problem with that. Like to use your analogy of you're walking to the top of the stairs, right? And they will basically plan out these are the stairs that I'm going to walk on... Maybe there's like 10 options of stairs and it's like I'm going to do this one then this one then this one then this one and then gets to the top before they start walking.”

0.52

Slight differences in skill are now leveraged to a historically unprecedented degree because AI and communication tools have removed the gaps that previously masked skill differences.

forecasthigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“It seems like we now live in a world where a very slight difference in skill is leveraged to a degree it's never been leveraged before... the people who were I already thought were like 10x are now 100x.”

0.52

If a decision is irrelevant to your P0 (top priority), it doesn't matter much what you decide—just do something and move on. If a decision hurts your P0 by distracting from it, the answer is automatically no.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Unidentified Speaker — $11B AI CEO: Why You Must Maximize Stress to Win [j6jgpb8kCR0]

“Um and based off of what your P0 is, does this help the P0? Does this hurt the P0? Or is this irrelevant to the P0? Why that's important is if it's irrelevant to the P0, it doesn't matter that much what decision you make. And so just do something and go. If it's negative to the P0, as in it like takes distraction away from the P0, the decision is no. Full stop.”

0.52

AI allows good engineers to work asynchronously without needing management or communication, shipping products solo on weekends that get seen by the whole team, breaking the constraint of communication/management.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Unidentified Speaker — $11B AI CEO: Why You Must Maximize Stress to Win [j6jgpb8kCR0]

“The second reason is you just don't need to communicate. Like you can just go and on the weekend on a Sunday, if you're really good engineer, you can just go build something and then just shoot it off into, you know, the general Slack and people can see what it is and so you don't even need those communication barriers, right?”

0.52

The best founders have built a scalable organizational machine that allows them to delegate execution, freeing them to focus exclusively on identifying and solving the company's primary bottleneck.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“what they're really good at is two things. One is they built a machine. And so because they've built a machine, they have time to only focus on what is the main bottleneck with the machine, right? And so it's those are two different skill sets, I think.”

0.52

Founders retain massive informational advantage over newly hired executives because they carry continuous context from day one, meaning the founder only needs to process the delta (change) from one day to the next, while new hires must absorb 10 years of accumulated context.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“The main advantage of being a founder of a company is if you are always on, which I tried to be, what you have is you have the context from day one, right? And if you're always on, your brain all your brain has to do is do the delta between day one and day two, right?... Whereas if you're a newcomer and you just come into a company, you don't have that. And so you're starting from zero and you have 10 years of context that you have to somehow absorb, right?”

0.52

Initial strategy to convince lawyers of Harvey's value was to personalize demos using their own recent work product, which dramatically increased engagement and made the AI capabilities tangible in a way generic examples could not.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“what I would do is I'd grab that and I'd say what's good about this argument? Could you argue against it?... you'd go from the lawyer like I used to have demos in the beginning where lawyers would literally be on their phone... And when I started saying hey you filed a brief last week and you know Apple V whatever case and you want to see how Harvey would analyze that brief... they were completely... watching the screen and read every single word that came out of Harvey.”

0.52

The founder wrote a paragraph-requirement rule where his chief of staff forces him to write a full paragraph justifying any meeting he wants to take, which makes it obvious when he's hedging or unsure—and this is an effective prioritization tool.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Unidentified Speaker — $11B AI CEO: Why You Must Maximize Stress to Win [j6jgpb8kCR0]

“When I get really bad at it, I ask my chief of staff to force me to write a paragraph about why I'm going to take a meeting. Full paragraph about why I should take a meeting. And it's so easy if you do it that way, because for 99% of meetings or events or whatever, you start writing the first sentence and you're like, I don't want to do this. And if you don't want to do this, and you're like, this is a waste of time, probably the meeting or the event is also a waste of time.”

0.52

Law students should focus on understanding clients and industries deeply rather than on technical legal skills (brief-writing, research), because AI will handle technical work while client understanding cannot be automated.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“if you can understand cut like your clients super well, and you understand an industry super well, you're going to do great. Like AI isn't going to massively impact you. If you think that your competitive advantage is you are the best at writing briefs, or you are the fastest at doing research, or something like that, that was never your competitive advantage.”

0.52

The act of frequently re-ranking daily tasks (clicking into a prioritization document and re-ordering items) increases metacognitive awareness about what you're actually doing and correlates with better performance and focus.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“every day I rank everything I need to do. And so, it's daily, and then I refresh it, and then I cross them out. And the re-ranking is I think really important because what I found is the more times I click into that doc that's called the list, and I re-rank something, the more I think about what I'm doing. Like meta thought about my thoughts. And that is really good.”

0.52

Company success is defined by re-earning position every 6 months with increasingly higher bar; the speaker wants to feel they've done everything possible to clear that bar.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“I really feel like you have to like re-earn your position every 6 months. And to me every 6 months that bar gets so much higher. And right now it I mean doubles or triples in in height. And I want to get to the point where I feel like we did everything we could every 6 months to get over that bar.”

0.52

Stress focuses people and increases decision quality, but only after going through the stress and resolving it; the stress itself is valuable because it reduces future stress for that category of decision.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“Yeah, massively. Um I think it's less it's less stress, it's going through stress and then that reduces it for the next thing... you just have to make tons of mistakes with things that you think are one-way doors to realize that not everything is a freaking one-way door, right?”

0.52

People with very prestigious backgrounds (never got a B, always succeeded) often cannot handle high-stress environments because they lack experience with failure recovery and will 'massively break' under repeated setbacks.

causalhigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“there are just so many folks that have these such prestigious backgrounds and they've never gotten a B in their entire lives, like ever, probably, right? And so they think everything is the end of the world, right? And you can't hire people like that cuz they will break. They will massively break because the reality is these companies are moving too fast.”

0.49

Despite law firms using AI tools like Harvey, legal fees are not decreasing because current tools do task automation rather than entire workflow automation, making it hard to quantify and pass through savings.

factualhigh valuespeaker onlynovelty 2/4durability 2/4· Winston Strawn

“I think that there's two reasons for this. One is the tools at this point do task automation, but they don't do like entire workflow automation. So it's hard to exactly say where the savings are, right?”

0.49

Data rooms may grow 20-50x larger in future M&A if agents create all contracts and negotiate with each other; human cognition cannot absorb this complexity alone, requiring agent+human teams.

forecasthigh valuespeaker onlynovelty 2/4durability 2/4· Winston Strategist

“Think about like a data room, what a data room is going to have in 10 years, if you've been using AI to create all the contracts. It's going to be 20 times larger, right? 50 times larger? It might get to a point, honestly, where we have agents that are basically creating all the contracts, negotiating with each other, and a human can't absorb all this data. Like you literally can't. Like it has to be an agent plus a human, otherwise you wouldn't be able to do the complexity of these transactions.”

0.48

The speaker has repeatedly experienced moments throughout life where they believed 'everything was over' (failed out of high school, poor LSAT score, company near-death), which created psychological resilience by normalizing recovery from perceived catastrophic failure.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“I've had a lot of times in my life when I thought everything was over. Like when I like got basically failed out of my first like high school I went to. I was like, this is over. This is done. I remember when like I took the LSAT for the first time... I was like, okay, well, that's over. But my point is you just have to get used to this of like it's over, it's over, it's over, it's over and then everything's fine.”

0.48

Principle 3 of running a company: Vision setting requires massive flexibility; the speaker has struggled with this, having created a long 5-year vision that was interpreted as immediate marching orders, disrupting product timelines.

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“The third thing I think is vision setting needs to have flexibility like massively... how do you vision set at the right level? That I still struggle with massively where sometimes... I wrote this like long doc and it was like the vision for the product for like five years or whatever a long time. And I think it was interpreted as like okay we should go build that right now.”

0.48

Successful founders maintain constant pain because they systematically neglect well-functioning parts of the organization and only focus on burning problems, which creates a perpetually stressed state.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“And I think that that's hard because that means you are going to be in pain 24/7. Because if something's going well at the company, you are not going to be working on it.”

0.48

The best founders have repeatedly ignored external pressure to make short-term improvements and instead focused on fixing the real long-term problem, vindicated 6-12+ months later despite intense criticism during the waiting period.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“I've found is a lot of the best founders that I look up to, they've had tons of moments where the entire outside world is like, 'Go do this thing. Go do this thing. You need to improve something.' And they just put blinders on and they completely ignore everyone. And they go do the hard thing that they know is actually going to fix the the problem long-term.”

0.48

In early 2024, Harvey attempted to acquire a company 10x their size at a similar valuation by securing a deal before financing was complete, using a payment-in-kind (PIK) debt structure that would have made the company vulnerable to ownership loss if debt obligations were unmet.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“We thought we we basically thought that there was going to be a way to like one shot ourselves into like building the company... The idea was basically we're going to buy this company that ended up it was 10 times bigger than us people wise. Um and about the same like we're going to trying to buy the company for the same a little bit higher valuation than we were valued at. We tried to buy it. We kind of like signed actually the deal um before we had the money.”

0.48

After failing to secure acquisition financing, the speaker initially believed the company was dead, but recovered within 24 hours to realize they needed to build through organic growth instead of shortcuts.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“when that happened and I remember just pulling it off and telling the other side we weren't going to do it and everything like that, I thought the company was over. I was just like, I don't think I don't think like we can build fast enough... And it forced us to basically go and like build the company. Like actually build the company.”

0.48

No one at Harvey has ever been fired for making too many mistakes; people are only let go after experiencing decision paralysis, inability to scale, fear of hiring better people, or other psychological breaks from stress.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“No one at the company ever has been let go because they made too many mistakes. It's never happened. I've I have personally never done it. I've never ever fired someone or anything like that because they made too many mistakes. They always broke.”

0.48

Some of the most productive people in the world are difficult to work with and poor at managing relationships, but companies often need to create special environments for them because their raw output is so exceptional.

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“Some of some of the best producing people on Earth are terrible communicating and terrible at working with others and a lot of what you sometimes have to do as a company is how do you create a situation for those folks to succeed because they can't manage up, you know, they don't look as prestigious, they maybe not don't have like the resume that everyone likes, they aren't as good at communication and they can be hard to work with but they're geniuses.”

0.48

Harvey balances hiring experienced executives with internal promotion because sometimes junior people with high raw talent will outpace senior hires within 2-3 years, making investment in slope more valuable than purchasing experience.

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“we hire a lot of like execs that have been there, done that, but we also promote a lot internally, right? Because sometimes folks are better than executives that I could have hired that were more senior... because they just have higher raw talent. And so it's better to invest in the junior um because the slope is just going to keep going and in 2 years, 3 years, they're going to be way higher than the senior exec that you thought about hiring.”

0.48

Deals will close faster (48 hours vs 3 weeks) as AI accelerates diligence and documentation, creating competitive pressure for speed that benefits early adopters.

forecasthigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“if you're a firm and you say, 'Hey, I can get this deal closed in 48 hours.' And another firm says, 'It'll take me 3 weeks.' If I'm a CEO and I want a deal to get done, I want the one that's done in 48 hours so it doesn't get leaked to the press, um so that you know, someone doesn't get cold feet or something, my stock price goes down, something like that, right?”

0.48

The founder wants to ensure Harvey doesn't collapse or fail because he never wants to return to life before the company; this drives his commitment and refusal to lose.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strategist

“And so I think I'm just extremely grateful for like the position I'm in and I don't want to lose that position. Like I just don't want to lose it. And I don't want to feel like I you know, kind of didn't put everything that I possibly could into it when I had the opportunity.”

0.48

Top M&A attorneys become valuable because they have personal relationships and knowledge of all the players in the market, enabling them to advise not just on what's legally possible but on what the other side actually wants.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“when you're doing a large M&A, a lot of the very good like top M&A attorneys, what they are good at is they actually know personally all the other folks, like all the players. And so, what they're ending up becoming is actually like deal advisors of I know that this is actually what they want.”

0.47

The speaker uses a live-editing, async-work hiring process (shared Google Doc with questions/projects) to assess both async work style and decision-making quality; excessive hedging in writing predicts decision paralysis in high-stakes situations.

normativehigh valuespeaker onlynovelty 2/4durability 3/4· Winston Strawn

“I go through like a Google Doc, and I do like a live editing process. Um so, I'll I'll come up with a Google Doc that has like a bunch of questions or like a project to do, and we'll do it live async... The other thing that it does is you can tell when some people like take too long. Like they'll write like a 20, you know, paragraph response to something. And you're like, you're just hedging constantly.”

0.45

Skill advantages in knowledge work compound similarly to sports: a small speed advantage in basketball compounds into blocking more shots, higher shooting percentages, and consistent wins.

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“I think it's the same thing that happens in sports, right? So like in sports being a little bit faster compounds, right? If you're a little bit faster than your opponent in basketball, right? You will just slightly get the ball in front of them every time.”

0.45

The speaker deliberately sacrificed short-term competitive ground on certain metrics (customer acquisition, speed) to optimize for long-term success (post-sales excellence, customer retention).

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“We've sacrificed a decent amount of ground to competitors because, you know, we really wanted to land customers and then keep them. And so we invested in a tremendous amount on like what happens post sales, right?”

0.45

The speaker did not find meaningful direction in life until founding Harvey at age 27+; this is the most fun and most painful experience of their life, and represents the difference between 'before' and 'after' in their life narrative.

factualhigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“I didn't really find my way for like 27 years before doing this company. And I definitely had some periods in my life of like dealing with pretty bad like mental health problems... This is the most fun I've ever had. Like it's not even close... there was a life before this and there was a life after.”

0.45

The founder prioritizes learning rate, resilience, and failure tolerance as hiring criteria over traditional credentials, because these qualities compound and are harder to fake than titles or backgrounds.

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strategist

“what I'm looking for a lot is are you able like what is your rate of learning? What is your Are you going to be okay with making mistakes? Can you have failures? Can you learn from them? What is your resilience? That I think is really important, too. And what is your company's resilience? Like that I think is something that a lot of companies are going to face pretty soon.”

0.45

Lawyer obligations to disclose AI use in client work are currently determined client-by-client and jurisdiction-by-jurisdiction, and have changed every 3-6 months from prohibition to conditional use to requirement.

factualhigh valuespeaker onlynovelty 1/4durability 2/4· Winston Strawn

“It completely depends on the client right now. A jurisdiction by jurisdiction and state by state, but right now mostly it is driven like the consensus of it is is driven by client by client... 2023 was don't use it at all or you have to tell me exactly what you're using it for... 2024 was you can use it by certain different types of matters... And then 2025 started being you need to use it and you need to tell me how you're using it and how you're saving me money.”

0.45

Many people who tried ChatGPT or early AI tools and experienced hallucination once have not revisited them for years, and remain skeptical despite massive capability improvements, because they don't track the pace of model advancement.

factualhigh valuespeaker onlynovelty 1/4durability 2/4· Winston Strategist

“they're still not using it because of that. Like I think there are still folks that have tried this like 3 years ago and they had a hallucination one time and they're like, 'Yeah, we're not going to do that.' I mean, we had this with Harvey where we had, you know, we've had so many folks that said we tried your product like 3 years ago, we didn't think it was good enough and we don't we're unsure if it's improved. And then they log in and they're like, 'I don't even remember trying the product 3 years ago.' Like this is so different.”

0.45

In 5 years, agents will execute most work; humans will specialize in reviewing agent outputs and catching mistakes—replicating what senior lawyers already do with junior lawyers.

forecasthigh valuespeaker onlynovelty 1/4durability 2/4· Winston Strategist

“My guess is we're going to get to a stage where agents are running a very large portion of this, and humans get really good at reviewing the outputs of agents and understanding where they might have made a mistake. This is also what senior lawyers are really good at already.”

0.43

Testing GPT-3 on Reddit legal advice questions (landlord-tenant law) produced outputs that 3 out of 3 landlord-tenant lawyers would send without edits in 86 out of 100 cases, creating the pivotal moment to start Harvey.

normativehigh valuespeaker onlynovelty 1/4durability 3/4· Winston Strawn

“We grabbed a bunch of landlord-tenant questions in California, and we ran this chain of thought prompt over them. And we basically gave it to three landlord-tenant lawyers. And we said, somebody asked this question... Would you send this? Like if you were the lawyer here, would you send this with zero edits? On 86 out of 100 of those questions, three out of three attorneys said yes.”

0.33

The speaker has filled every role at the company at some point and then hired someone to take that role, which gives them confidence about what's happening in each area and grounds their decision-making.

factualestablishednovelty 0/4durability 3/4· Winston Strawn

“I've also done every part of the company. Like I've played every single role at the company at one point. Like I've run everything at some point and then hired somebody to take that.”

0.30

AI can translate domain-specific problems into other domains for comparison (e.g., fund formation into engineering analogies), enabling better understanding and removing one barrier to idea sharing.

factualspeaker onlynovelty 1/4durability 3/4· Winston Strawn

“Something I do a lot now is the models have gotten so good that I can use a really complex like something that's really complex in legal, right? Like multi-jurisdictional fund formation and you have to do like the how you do tax exemptions and all these things. I can just say, can you give me an example in engineering that's similar to this?”

0.24

The co-founder has deep AI research background, having been in the first Google Brain cohort and worked at Meta, which provided credibility and technical depth for building AI-native legal product.

factualspeaker onlynovelty 0/4durability 4/4· Winston Strawn

“I met my co-founder. Um yeah... And he had a background in AI research. So, he was one of the first folks at the at Google Brain. He was the first Google Brain class, actually. And then he worked at Meta after that.”

0.23

The speaker refused early advice to hire senior executives too quickly, then waited too long before hiring them, missing a 6-month window when early hiring would have been optimal.

factualspeaker onlynovelty 0/4durability 3/4· Winston Strawn

“I was told to hire senior executives very very early on and I refused and then I refused too long. I refused like I'd say about 6 months too long. But if I would have taken the advice when it was given, I think it would have been a huge mistake.”

0.19

Harvey AI applies foundation models (primarily GPT-based) to legal work by building a system that tracks model progress and applies it to legal workflows, enabling lawyers to work more efficiently and changing how legal departments operate.

factualspeaker onlynovelty 0/4durability 2/4· Winston Strawn

“What we're doing is building a system for lawyers to use all of these models to do their work in a much better way and much more streamlined way. That's an incredibly broad explanation for what the company does. Another way to think about it is we are basically tracking the progress of all of these models in the system and applying it to an industry, right?”

0.17

OpenAI (Sam Altman and Jason Quan) became Harvey's first investor after receiving a cold email with results of the Reddit legal advice experiment, validating the product concept before building at scale.

factualspeaker onlynovelty 0/4durability 2/4· Winston Strawn

“We basically said, 'Hmm, we're using GPT-3 to do this. Let's just cold email Sam Sam Altman and the general counsel at the time whose name is Jason Quan... And we emailed them the results... They got back to us like pretty fast. We met with Jason first... Then we met with the rest of the leadership team on actually the 4th of July and we pitched them our idea for the company and we raised money after that.”