Feb 14, 2025 · 49m · no-priors
No Priors Ep. 101 | With Harvey CEO and Co-Founder Winston Weinberg
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of No Priors, Harvey co-founder and CEO Winston Weinberg joins host Sarah Guo to discuss building the leading AI platform for legal and professional services. He details Harvey's founding story, product architecture, go-to-market strategy with elite law firms, and the broader transformation of professional services through reasoning models and domain-embedded AI.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 24.7% of the talking time here. How this is scored →
The hosts as informed peer 5.6 Guest teaching 3.0 Guest disagreement 1.1 The hosts pushing back 1.2
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 24:24 Winston insists law firms do work at a loss
Winston firmly pushes back against the conventional belief that law firms are universally high-margin across every task, detailing private equity loss-leader practices.
Hardest push from the hosts ▶ 24:23 Sarah challenges law firm loss claimSarah immediately interrupts and refuses Winston's framing that elite law firms take losses on client work, stating 'It doesn't feel like that.'
Biggest teaching moment ▶ 8:40 Winston explains benchmark inadequacyWinston educates Sarah on why standard AI benchmarks fail to measure real-world legal competence, explaining why high-level domain experts are required for evaluation.
The host holds their own ▶ 25:01 Sarah synthesizes investment banking parallelSarah demonstrates deep market expertise by contextualizing Winston's point into a broader financial analogy of loss-leader advisory work in investment banking.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Platform Strategy: Expanding and Collapsing Workflows | 5 | 3 | 1 | 1 | Sarah sets the context as an early investor and asks Winston to elaborate on Harvey's product scope. Winston outlines his core product philosophy of expanding into granular vertical workflows and then collapsing them into a unified, simple UI. | |
| AI Patterns and Domain Expert Evaluation | 5 | 4 | 1 | 1 | Sarah inquires how Harvey integrates domain expertise with research and engineering. Winston explains their 'AI patterns' architecture and reveals why standard public LLM benchmarks are virtually useless for real enterprise legal evaluation. | |
| Complex Workflows and Minimum Viable Quality | 6 | 3 | 1 | 1 | Sarah asks how Harvey built trust with conservative, high-billing lawyers given early hallucination risks. Winston details the distinction between minimum viable quality in productivity tools (show citations) versus end-to-end specialized agents. | |
| Enterprise Go-to-Market with Prestigious Institutions | 6 | 4 | 1 | 2 | Sarah questions why Harvey defied traditional SaaS bottom-up/mid-market wisdom to sell directly to century-old elite firms. Winston explains that winning industry trust and critical data partnerships requires conquering the hardest tier first. | |
| Simplifying Enterprise UX and Orchestration | 6 | 2 | 1 | 1 | Sarah and Winston discuss how LLM orchestration enables radical simplification of enterprise UX. Winston explains how complex multi-step share purchase agreement workflows can be collapsed down into simple conversational prompts. | |
| Impact on Legal Careers: Task vs. Job Displacement | 5 | 3 | 1 | 1 | Sarah asks about user sentiment and automation fears. Winston explains the shift from panic to adoption, stressing that AI represents task displacement (repetitive doc review) rather than job displacement, accelerating associate career progression. | |
| Career Advice and the Evolution of Law Firm Business Models | 6 | 4 | 2 | 2 | Sarah asks about the future of the billable hour and law firm business models. When Winston claims law firms perform substantial work at a loss, Sarah pushes back sceptically, prompting Winston to clarify loss-leader dynamics in private equity deals. | |
| Reasoning Models and Cross-Industry Expansion | 5 | 3 | 1 | 1 | Sarah inquires how reasoning models and test-time compute scaling affect Harvey's roadmap. Winston explains how reasoning models unlock multi-step compliance chains and enable parallel expansion into tax and audit diligence. | |
| Company Culture, Hiring for Agency, and Founder Scaling | 6 | 2 | 1 | 1 | Sarah asks how engineering hires learn domain context and what mistakes Winston made as a first-time founder. Winston explains why hiring for raw agency and adaptability trumps traditional domain experience in fast-moving AI. | |
| Work Ethic, Imposter Syndrome, and Building Intuition | 6 | 2 | 1 | 1 | Sarah probes Winston's extreme work ethic and how he navigates imposter syndrome without a research science background. Winston cites Kobe Bryant's 'job's not finished' mentality and explains that absorbing intuition from world-class peers builds domain confidence. | |
| Opportunities in Application AI and Surprising Management Talent | 6 | 3 | 1 | 1 | Sarah and Winston discuss heuristics for application-layer founders. Winston shares his 'price per token' framework, advises founders to observe real industry problems outside Silicon Valley, and highlights the value of promoting internally for ownership. |
Statements from this episode (23)
Guo: Harvey has raised over $500M from OpenAI, Sequoia, Kleiner, and others
“They've now raised more than five hundred million dollars from investors such as OpenAI, Sequoia, Kleiner Perkins, GV, Bloodgill, and me.”
Weinberg: AI founders must aim to transform entire industries
“If you are using these tools and you don't think that you can take basically AI and apply to X industry and transform the entire industry, I don't think you're thinking ambitiously enough, right?”
Weinberg: AI platforms must expand modular workflows and collapse UI complexity
“These models can't just one shot all of these really complex legal tasks or in these other domains like tax and provide other professional services. And so what you have to do is you have to build a platform that is kind of constantly expanding and constantly …”
Weinberg: Intuition for AI progress comes from hammering models through failures
“It seems like there was this large gap between
I log into ChatGPT or I just use GPT-IV from an API or whatever it is, and I try something a couple times versus I'm going to sit there and just hammer on this until it works, right?
and I think if you do that …”
Weinberg: Enterprise AI requires pairing broad productivity tools with chained vertical workflows
“You need to build productivity tools, and what I mean by this is
Things that are useful for the highest amount of seats, right?
And then you also need to build things that are streamlined vertical workflow from start to finish, right?
And what you can do is…”
Weinberg: Standard AI benchmarks are useless for evaluating legal AI
“Most benchmarks are completely useless for us, right? And so we'll get a model, you know, someone will give us early access to a model and they'll say it's way better on all of these benchmarks and we'll respond. It actually isn't like, it's not used as useful…”
Weinberg: Vertical AI evaluation requires senior domain experts, not junior staff
“And the reality is you have to hire very good lawyers who can actually evaluate these systems, and the same within tax and these other areas, and they can't be too junior, because if they were too junior and they were able to eval it, they would be senior, rig…”
Weinberg: Productivity AI needs lower quality thresholds if work is verifiable
“On the productivity side, the minimum viable quality of that output can be lower because you're selling seats, And at the end of the day, there are multiple people reviewing it, right? And so what you want to do in that state is have just show your work, right…”
Weinberg: Evaluating scoped AI workflows is easier than open-ended copilots
“It's much more scoped, and evaluation is much easier, right? Like, how do you do evaluation on copilot or enterprise GPT, et cetera? That is very difficult. It is much easier to do evaluation if you can do evaluation at each step and make sure that recursively…”
Weinberg: Earning enterprise trust requires targeting the hardest customers first
“We found that no matter what, you eventually have to get the trust of the industry. And the best way to do that is to actually go after the hardest people first.”
Weinberg: The fundamental UI for professional services is just email
“And if you actually think about how the UI for professional services is just email, like that's it.”
Guo: Harvey serves over 250 clients and generates $50M+ ARR
“You guys now serve, you know, more than 250 clients, more than fifty million in ARR.”
Weinberg: AI will allow junior lawyers to do strategic work earlier
“And so what I think will end up happening is the timeline will compress. So you will start being able to actually do the high level strategic work and interact with clients, which is what people really want to do earlier on in your career.”
Weinberg: AI drives task displacement rather than job displacement in law
“It is not job. Displacement is task displacement. And I think that's a super important distinction because getting rid of those tasks does not mean the legal industry falls apart. It will evolve.”
Weinberg: Automated legal tasks will shift to fixed-fee pricing
“I don't think the billable hour is going to just completely disappear. I think what is going to happen is a lot of these tasks that are going to get, you know, AI can automate a lot of that task with a lawyer in the loop. Those tasks will end up being kind of …”
Weinberg: Harvey is helping law firms build and sell specialized software
“We were starting to work with firms and basically take, you know, the special things that they do and the way that they practice law and turn it into their specialized system in Harvey. And then they go and sell that to their clients.”
Weinberg: Law firms perform work at a loss for PE clients
“There is so much legal work that law firms actually do at a loss and they do it at a loss... There's a lot of legal work that they do at a loss in order to get the big deal. So, you know, a lot of law firms will do work for private equity and they will do that…”
Weinberg: Legal AI workflows transfer effectively to tax and finance
“There's so many areas where legal is kind of the tip of the spear and then you can actually just kind of paralyze the same things that you're building and tweak them. And it works really well in other industries.”
Weinberg: Founders should do every role themselves before hiring for it
“One of the lessons that I definitely want to keep is I do think you should do every single role For a certain amount of time before you hire for it. Almost all of my mishires were because I did not understand what that role was.”
Weinberg: Spending time with exceptional people is undervalued compared to asking questions
“I think that it is very overvalued how much value you get from asking people that you really respect, very specific questions, I think is overvalued. I think spending time with people that are incredible at whatever they do is undervalued. And I think it's bec…”
Weinberg: Vertical AI founders should evaluate the 'price per token'
“So one thing that I look at is you can look at this by industry and you can look at it by task, which is how expensive is the token? And I, I've come up with like different ways to call this, but basically if you look at something like that share purchase agre…”
Weinberg: Senior employees protect positions, making critical feedback seekers rare leadership signals
“There are a few folks that will come up to me and basically just not ask for positive feedback and just say, what did I get? Like, what do I need to improve? That is a massive signal because I think that something that happens as you get more senior in your ca…”
Weinberg: AI will produce specialized ChatGPT-level breakthroughs in coding and medicine
“I think we will end up seeing work completion done in very sophisticated manners, like in medicine and in coding, like a lot in coding and a lot also in, you know, other areas too, that will kind of bring us back to the chat GBT moment, but it will be speciali…”