Aug 9, 2025 · 1h 35m · lennys-podcast
Inside ChatGPT: The fastest growing product in history | Nick Turley (OpenAI)
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 in-depth interview, Lenny Rachitsky speaks with Nick Turley, Head of ChatGPT at OpenAI, exploring the 10-day origin story, scaling dynamics, GPT-5 rollout, and product philosophy behind the fastest-growing consumer application in history.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Lenny holds 21.6% of the talking time here. How this is scored →
Lenny as informed peer 3.5 Guest teaching 4.3 Guest disagreement 0.7 Lenny pushing back 0.8
speaking balance: gold is Lenny, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 35:00 Nick rejects the permanent chat interface paradigm
Nick pushes back against the widely held industry assumption that turn-by-turn chat is the ultimate AI interface, calling it limiting and borderline dystopian.
Hardest push from Lenny ▶ 28:49 Lenny challenges Nick on what sits on top of the modelLenny pushes past vague high-level assertions to press Nick on what OpenAI's product team actually builds that creates value beyond the underlying intelligence.
Biggest teaching moment ▶ 29:18 Nick explains why the model is the productNick reframes the host's question by breaking down how modern AI product development eliminates the distinction between model training and feature design.
Lenny holds their own ▶ 40:35 Lenny identifies the Van Westendorp pricing surveyLenny immediately cites the formal Van Westendorp pricing methodology and its Silicon Valley pedigree when Nick only remembers asking four generic questions on Discord.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Lenny as informed peer | Guest teaching | Guest disagreement | Lenny pushing back | Why |
|---|---|---|---|---|---|---|
| Episode Preview and Highlights | 0 | 0 | 0 | 0 | Introductory preview and show setup by the host summarizing Nick Turley's background. As a monologue segment, all host-interaction metrics are zeroed out. | |
| Sponsor Message: Orcus Conductor | 0 | 0 | 0 | 0 | Sponsor segment featuring Orcus Conductor and a brief scripted interview with Christina Cacioppo of Vanta. Standard promotional dialogue with zero debate or technical pushback. | |
| Introducing GPT-5: Smarts, Utility, and Speed | 3 | 5 | 1 | 1 | Lenny prompts Nick on the launch of GPT-5, and Nick educates the audience on why dynamic reasoning, coding benchmarks, and output taste represent a generational leap over previous models. | |
| The Super Assistant Vision and User Agency | 4 | 5 | 1 | 1 | Lenny synthesizes the assistant paradigm, prompting Nick to outline OpenAI's design philosophy around keeping humans in control rather than building autonomous systems that replace users. | |
| The 10-Day Origin Story of ChatGPT | 3 | 6 | 1 | 0 | Nick reveals the internal history of ChatGPT originating as a 10-day sprint with volunteer engineers to generate real user data distributions before deprecation. | |
| Managing Scale, Velocity, and Empirical Product Observation | 3 | 4 | 0 | 1 | Lenny inquires how Nick handles managing a 700M user product, and Nick explains why empirical observation of emergent AI behavior is foundational to product strategy. | |
| Urgency, Resting Heart Rate, and Maximal Acceleration | 4 | 5 | 1 | 1 | Lenny brings up internal OpenAI insider terms like maximal acceleration, prompting Nick to explain how forcing questions accelerates delivery while deliberate process protects safety. | |
| Retention Dynamics and the Smiling Curve | 5 | 4 | 0 | 1 | Lenny demonstrates strong product expertise by introducing retention numbers and the smiling curve concept, which Nick validates while discussing the learning curve of AI delegation. | |
| Treating the Model as the Product | 5 | 6 | 2 | 3 | Lenny directly challenges Nick on what product layers OpenAI builds beyond raw model intelligence. Nick educates by explaining that the model itself must be treated and iterated on as the product. | |
| Beyond Chat: Rethinking the Interface of AI | 4 | 6 | 3 | 2 | When Lenny references Kevin Weill's defense of the chat format, Nick pushes back against conventional wisdom, arguing natural language will persist but turn-by-turn chat is limiting and dystopian. | |
| Accidental Decisions: Naming, Monetization, and $20 Pricing | 6 | 5 | 0 | 1 | Nick explains how they frantically priced Plus at $20 using a four-question Google Form on Discord, and Lenny instantly identifies the academic method as the Van Westendorp Price Sensitivity Meter. | |
| Enterprise Expansion and Multi-Front Product Strategy | 4 | 5 | 1 | 1 | Lenny asks how OpenAI manages B2B enterprise alongside consumer and API. Nick uses a Disney IP analogy to explain working backwards from core model capabilities. | |
| Out-of-Product Discovery and Future Interfaces | 4 | 5 | 1 | 1 | Lenny contrasts horizontal tool adoption struggles with ChatGPT's explosion, and Nick explains how removing the waitlist turned public viral comments into an out-of-product discovery engine. | |
| Sponsor Message: PostHog All-In-One Product Platform | 2 | 3 | 0 | 0 | PostHog sponsor message followed by a brief discussion on how OpenAI's data science team uses conversation classifiers to discover organic user behaviors. | |
| Mitigating Sycophancy, Aligning Incentives, and High-Stakes AI | 4 | 5 | 1 | 1 | Lenny and Nick discuss the sycophancy bug, with Nick explaining OpenAI's duty to build robust frameworks for high-stakes medical and relationship advice rather than reflexively blocking sensitive use cases. | |
| Lean Teams, Barrels vs. Ammunition, and Research-Product Culture | 4 | 5 | 0 | 0 | Nick explains OpenAI's lean staffing philosophy and applies Keith Rabois' barrels versus ammunition framework to hire cross-functional leaders who can unblock shipping. | |
| First-Principles Product Craft and Shipping Raw Features | 3 | 5 | 1 | 1 | Nick breaks down first-principles thinking in product craft, using OpenAI's much-criticized model dropdown chooser as an intentional trade-off of polish in favor of raw user learning. | |
| Speed Over Premature Polish and the Critical Role of Evals | 5 | 5 | 1 | 1 | Nick explains why velocity is required to uncover emergent failure modes, and Lenny connects this to evals becoming the mandatory lingua franca for AI product managers. | |
| ChatGPT as a Web Traffic Engine and the Future of Search | 5 | 4 | 2 | 1 | Lenny shares that ChatGPT drives more traffic to his newsletter than Twitter. Nick rejects trending jargon like AEO and GEO, explaining models should simply align with authentic user intent. | |
| The Evolution and Potential of Custom GPTs and App Ecosystems | 3 | 4 | 0 | 0 | Nick reviews the status of Custom GPTs and reflects on his undergraduate philosophy background analyzing whether rational people can disagree. | |
| Joining OpenAI and Core Career Principles | 3 | 4 | 0 | 0 | Nick describes his path into OpenAI starting with fixing the office blinds, offering career guidance on prioritizing curiosity and high-caliber peers over short-term compensation. | |
| Lightning Round: Sci-Fi, Life Mottos, and Jazz Improvisation | 2 | 3 | 0 | 0 | Lightning round covering sci-fi inspirations like Her and Severance, favorite books, life mottos, and how jazz band improvisation mirrors high-velocity software development. |
Statements from this episode (41)
Turley: ChatGPT has 700 million weekly active users
“If you're the average ChatGPT user, and we have, you know, seven hundred million of them this week, we you know, you've probably been on GPT-V for, you know, a while.”
Turley: GPT-5 achieves state-of-the-art results on math and reasoning benchmarks
“One way to look at that is academic benchmarks on many of the standard ones whether or not it's math or reasoning or, you know, just raw intelligence, this model is state of the art.”
Turley: GPT-5 dynamically chooses when to reason without manual user prompting
“It thinks too, just like O three did, but you don't have to manually, you know, tell it to do that. It'll just dynamically decide to think when it needs to. And when it doesn't need to think, it just responds instantly.”
ChatGPT's original codebase was a hackathon project called 'SA server'
“In fact, the code base that we use is called SA server. It was supposed to be a hackathon code base.”
Turley: OpenAI developer platform has around four million developers
“In fact, it's grown up and now it's the OpenAI platform with, I don't know, four million developers, I think.”
AI product builders must ship to learn rather than reasoning upfront
“This is a pattern with the, I think, where, you know, you really have to ship to understand what is even possible and what people want. Rather than being able to reason about that a priori.”
Turley: OpenAI initially planned to shut down ChatGPT after holiday testing
“We shipped it right before the holiday thinking we would sort of come back and get the data and then wind it down.”
Turley: ChatGPT will hit one billion users soon
“We're gonna hit a billion users soon, and you kind of have to begin acting in a way that is appropriate to that scale.”
Turley: OpenAI shipped ChatGPT within 10 days of deciding to build it
“When we decided to do it, you know, we had been prototyping for so long, and I was just like, you know, in 10 days we're going to ship this thing, and, you know, we did, so that was like maybe a moment in time thing where I just really wanted to make sure that…”
Turley: AI companies must separate product velocity from rigorous model safety
“I think it's been really important to separate out, you know, the product development velocity, which has to be super high, from, okay, for things like frontier models, there actually needs to be a rigorous process where you red team, you work on the system ca…”
Turley: OpenAI does not care how much time users spend on ChatGPT
“You know, we don't care at all how much time you spend on the product. You know, in fact, our incentive is just to solve your problem, and, you know, if you really like the product, you'll subscribe, but, you know, there's, No incentive to keep you in the prod…”
Turley: ChatGPT user retention has improved over time across cohorts
“And it's been incredible to not just see strong retention numbers, but to see, you know, an improvement in retention over time. Even as our cohorts become, you know less of an early adopter and more, you know, the average person, so.”
Turley: Delegating tasks to AI is unnatural for most people
“I think people are actually just getting used to this technology in, like, a really interesting way where I find, and this is why the product needs to evolve too, that this idea of delegating to an AI, it's not natural to most people. It's not like you're goin…”
Turley: There is no distinction between the AI model and the product
“The one thing we've learned with ChatGPT is that there really is no distinction between the model and the product, like the model is the product and therefore you need to iterate on it like a product.”
Turley: Pre-ChatGPT, OpenAI shipped models like hardware instead of software
“Treating the model as a product was not a thing before ChatGPT, because we would ship it more like hardware, where, you know, there'd be a release like GPT-III, and then we would start working on GPT-IV, and these weird giant Big spend R&D projects that would …”
Turley: Turn-by-turn chat interactions are a limiting paradigm for AI
“So I think natural language is here to stay, but this idea that it has to be a turn by turn chat interaction, I think is really limiting.”
Turley: GPT-5 is amazing at building front-end applications
“GPT-Five is amazing at making great front end applications.”
Turley predicts AI systems will dynamically render their own UIs
“I don't see a reason why you wouldn't have, you know, AIs that, you know, can render their own UI in some way.”
Turley: ChatGPT was renamed from 'Chat with GPT-3.5' the night before launch
“We went from Chat with GPT, 3.5 to ChatGPT the night before,”
Turley: GPT-3.5 was in OpenAI's API for six months before ChatGPT
“The GPT. 3.5 model was in our API for, you know, at least six months prior to that. I think anyone could have built something like this.”
Turley: OpenAI created ChatGPT subscriptions solely to throttle excess demand
“The birth of that was just to turn away demand originally. Like, it was not like, you know, we brainstormed, oh, what is the best monetization model for AI? It was really, what is, what monetization model has, or what mechanism would allow us to turn away peop…”
Turley: OpenAI priced ChatGPT at $20 using a four-question Discord survey
“So what I did do is ship a Google form to Discord with like, I think the four questions you're supposed to ask on how to price something. Yeah, exactly. It literally had those four questions, and I remember distinctly, A, you know, I got a price back and that'…”
ChatGPT organically reached 90% of Fortune 500 companies early on
“We were pretty quickly in, you know, organically in, like, 90% of Fortune 500 companies”
OpenAI reaches 5 million ChatGPT business subscribers
“We just hit five million business subscribers up from three, I think a month or two ago.”
Turley: Traditional PM frameworks fail when applied to raw AI capabilities
“One is sort of working backwards from the model capabilities, and that is much more than science where I think you really need to look at what tech do we have available and what is like the most awesome way to product productize it. And if you apply to some so…”
Turley: ChatGPT was the first OpenAI release without a waitlist
“Actually, you know, we were debating waitlist, no waitlist because we really knew we couldn't scale the engineering systems. And, you know, the fact that there was No waitlist, which no open AI release had worked like that before.”
Turley: ChatGPT is AI's MS-DOS and 'Windows' is not built yet
“I still feel like ChatGPT feels a little bit like MS-DOS. We haven't built Windows yet, and it will be obvious once we do, but, you know, there, there's something that feels a little bit like, imagine MS-DOS, like, gone viral, and you were just trying to, like…”
OpenAI uses automated classifiers to track ChatGPT use cases without reading chats
“We have conversation classifiers that without, you know, us having to look at the conversations, allow us to kind of figure out what are people talking about, what use cases are taking off, et cetera.”
Turley: GPT-5 improves on sycophancy benchmarks over previous releases
“GPT five is an improvement, which is really exciting for me, but we have more work from there.”
Turley: Big tech avoids sensitive use cases, OpenAI leans into them
“I think that's what most tech companies do when they hit a certain skill. They run away from these use cases. And I think it's a lost opportunity to help people. So we want to run towards these use cases by making the model behavior really, really great.”
Turley: AI features that do not scale with model intelligence should not ship
“Like if, you know, we're shipping a feature and it doesn't get two X better as the model gets two X smarter. It's probably not a feature we should be shipping.”
Turley: OpenAI operates ChatGPT with a remarkably lean team
“So we've always had this tendency to run relatively lean. So it is a small team that is running chat GPT. I take inspiration from WhatsApp where like, you know, it was a very small team running a very global scope product.”
Turley: Saturated benchmarks mean shipping is the only way to find model failures
“The benchmarks are increasingly saturated. So really you need real world scenarios where your product or model is not actually doing the thing it was supposed to do. And the only way you get that is by shipping because you get back to sort of use case distribu…”
Turley: Evals are the lingua franca between product managers and AI researchers
“I was like, wow, this might be the lingua franca of how to communicate what the product should be doing. To people who do AI research. And that really clicked for me. And at the end of the day, it's not that different from The wisdom of you ought to articulate…”
Turley: Optimizing content for AI models mirrors appealing to human users
“The way to appeal to an AI model ideally is the same way that you would appeal to a real user because the model is supposed to proxy the interest of the user and nothing else.”
Turley: Consumer custom GPTs launched before models had sufficient tooling
“They're kind of ahead of their time in the sense that we built that kind of concept before you could really build very differentiated things, at least in the consumer space. You know you're like learning GPT is going to be pretty similar to what the model coul…”
Turley: OpenAI is seeing substantial enterprise adoption of custom GPTs
“Different in the enterprise, by the way, we're seeing a ton of adoption of GPTs there because just every single company has very bespoke business processes and problems, et cetera. And it's a really, really useful tool there. They also have unique data that th…”
Turley: Billion-user ChatGPT will enable new businesses via built-in distribution
“Like, I think there really is a world where, you know, as this thing hits billion user scale, it can get you distribution. It can get you know, started on making something in the same way that people built on the internet. And, you know, there was entirely new…”
Curiosity matters far more than prior ML experience for AI product hires
“For product and engineering and design people and, you know, those kinds of functions, I actually think that if you are just curious about the stuff works, it doesn't matter at all if you've never done it before. In fact, if you were to filter for people who'v…”
Turley: OpenAI's pace leaves no time to track third-party API products
“Sometimes, like, you know, API developers ask me, it's like, hey, are you, like, you know, going to copy all of our products? There is, like, I actually just do not have time to follow up, you know, what's going on outside of OpenAI, because the pace here is, …”
Turley: GPT-5 will be free and default, with legacy models for Pro
“Just use the product. You don't even have to pay. Should be your default model starting tomorrow. And just use it and don't think about models anymore. Unless you want to and you're a pro user, in which case you get all the old models, so rest assured.”