Jul 24, 2025 · 1h 10m · in-depth
What Braintrust got right about product-market fit | Ankur Goyal (Founder and CEO)
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
Braintrust founder and CEO Ankur Goyal joins Brett Berson to break down the realities of building AI infrastructure, detailing the transition from the trap of weak product-market fit to genuine customer pull, the necessity of extreme engineering paranoia, and the cultural frameworks required to build elite technical teams.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Brett holds 9.7% of the talking time here. How this is scored →
Brett as informed peer 4.8 Guest teaching 4.8 Guest disagreement 1.4 Brett pushing back 1.7
speaking balance: gold is Brett, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 34:55 Dying in obscurity vs overhype
Ankur firmly rejects conventional startup caution around launching early, declaring that dying in obscurity is unequivocally worse than being overhyped and failing.
Hardest push from Brett ▶ 1:00:04 Encoding personal quirks into cultureBrett challenges Ankur on whether his strict company rules are just an imposition of personal working preferences, pressing him to explain how he separates personal habits from company policy.
Biggest teaching moment ▶ 39:20 Relational databases failing at full-text searchAnkur breaks down the exact technical limitations of existing OLAP and relational databases when indexing 50KB AI span rows, explaining why Tantivi was the only viable search engine for Notion's scale.
Brett holds their own ▶ 19:34 The definition of weak product-market fitBrett crystallizes Ankur's early startup struggles into the framework of 'weak product-market fit', demonstrating deep venture expertise on how non-repeatable revenue leads founders into prolonged traps.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Brett as informed peer | Guest teaching | Guest disagreement | Brett pushing back | Why |
|---|---|---|---|---|---|---|
| Ankur Goyal's Early Path: From Medicine to MemSQL | 4 | 2 | 1 | 1 | Brett establishes personal rapport by recalling their early meeting at MemSQL when it was running out of an apartment. Ankur shares his origin story transitioning from wanting to be a doctor to dropping out of CMU to join MemSQL. | |
| Foundational Lessons from MemSQL: Paranoia, Responsibility, and User Alignment | 4 | 5 | 1 | 1 | Brett asks Ankur to reflect on the core lessons from his time at MemSQL. Ankur delivers a masterclass on engineering paranoia, high software quality, and non-transactional recruiting. | |
| Engineering for Quality: Diligence, Edge Cases, and User Trust | 5 | 5 | 1 | 3 | Brett pushes with crisp probing questions asking how to build high-quality software and why that level of paranoia is critical. Ankur details the mindset of senior engineers catching edge-case 504 errors on Friday evenings. | |
| Mastering Non-Transactional Technical Recruiting and Interview Rigor | 5 | 5 | 2 | 2 | Brett asks how to teach someone to recruit beyond platitudes. Ankur explains the counterintuitive dynamic of hard, rigorous interviews even for trusted candidates to avoid false positives. | |
| The Founding of Empira and Early Entrepreneurial Misconceptions | 4 | 4 | 2 | 1 | Ankur reflects openly on the mistakes he made founding Impira, such as assuming easy fundraising meant product-market fit or that technical difficulty equaled market desire. | |
| The Trap of Weak Product-Market Fit and Lack of Repeatability | 6 | 4 | 1 | 2 | Brett introduces the conceptual framing of 'weak product-market fit' to describe the trap of having paying customers without repeatability. Ankur strongly agrees, remarking that he would rather have zero customers than unrepeatable ones. | |
| The Inception of Braintrust: Addressing the AI Evals Dilemma | 4 | 6 | 2 | 1 | Ankur gives an in-depth breakdown of how repeating the same eval infrastructure build at Impira and Figma led to incubating Braintrust alongside Elad Gil. He describes running open-ended interviews with early AI teams like Zapier and Coda. | |
| The Strategic Power of Default Skepticism in Startup Execution | 6 | 4 | 2 | 2 | Brett highlights the paradox of Ankur being default-skeptical rather than an unshakeable visionary. Ankur explains how skepticism protects against market delusions while conviction drives engineering execution. | |
| Unmistakable Product-Market Fit: Organic Pull and Unforced Demand | 5 | 6 | 1 | 1 | Ankur contrasts the bespoke, high-effort sales demos of earlier companies with the organic inbound pull at Braintrust, where prospects pointed at generic screens demanding the product immediately. | |
| Community Dynamics: Shared Problem Spaces and Peer Evangelism | 6 | 4 | 1 | 2 | Brett synthesizes Ankur's insights about market urgency and repeatability. Ankur compares Braintrust's community dynamic to dbt, where practitioners gather organically around an unsolved operational problem. | |
| Public Launch Strategy: Embracing Public Visibility Over Obscurity | 4 | 4 | 2 | 1 | Ankur challenges the conventional founder fear of launching an unpolished product, arguing that dying in obscurity is vastly worse than being overhyped and failing. | |
| Engineering Brainstorm: Purpose-Built Search Infrastructure for Notion | 5 | 7 | 2 | 2 | Ankur explains in technical detail why relational and OLAP engines failed at full-text search for 50KB AI spans, prompting his team to build Brainstorm on Tantivi in two and a half months to support Notion. | |
| Architecting for AI Volatility: Balancing Durable Bets and Ephemeral Trends | 5 | 6 | 1 | 2 | Brett asks how to build durable infrastructure amid fast-moving AI hype cycles. Ankur distinguishes permanent architectural pillars (stateful prompt playgrounds, high-volume span indexing) from ephemeral features (agent tool pricing quirks). | |
| Maintaining ICP Focus While Scaling to Traditional Enterprise Workflows | 4 | 5 | 1 | 1 | Ankur outlines how Braintrust stayed disciplined on their core engineering ICP while letting traditional enterprises eventually adopt modern AI workflows, adapting to enterprise-specific needs like Azure and PDF support without losing focus. | |
| Authentic Customer Relationships as the Catalyst for Brand Trust | 5 | 4 | 1 | 2 | Brett asks about the mechanics behind leading AI founders championing Braintrust on social media. Ankur emphasizes that brand evangelism cannot be manufactured and stems from authentic peer collaboration and shared problem-solving. | |
| Articulating Strategic Bets and Designing High-Autonomy Company Culture | 4 | 5 | 2 | 1 | Ankur discusses establishing clear cultural boundaries at Braintrust, including holding only one company meeting per week, cutting off meetings at noon, and filtering out candidates who need extensive process. | |
| Navigating Cultural Boundaries, Personal Habits, and Management Misfires | 6 | 5 | 2 | 4 | Brett probes on how a founder decides which personal quirks become company culture versus individual choices. Brett also presses Ankur to define what 'misfiring' means in startup management. | |
| Defining Product-Market Fit: Uncontrollable Demand and Inelastic Usage | 5 | 5 | 1 | 1 | Ankur defines true product-market fit as uncontrollable, inelastic customer demand where users flood the infrastructure. He concludes by crediting Alana Goyal and Elad Gil for teaching him counterintuitive startup principles. |
Statements from this episode (28)
Taking a product from 95% to 100% quality is exceptionally challenging
“It's pretty easy, and now I would say even like easier to build a prototype, but the craft of actually taking something from 95% to 99% or a hundred percent is very challenging and not something they teach you in school.”
Recruiting elite engineers requires a multi-year, non-transactional process
“Recruiting really, really good engineers is not a transactional process. I think that sounds very obvious to anyone who's tried or read about recruiting, but until you experience it, it's very hard. We sometimes spend years recruiting someone”
Figma is exceptionally good at recruiting talent
“For example, I was at Figma recently, I think they're really exceptional at it.”
Building high-quality software requires assuming new code is broken
“I think you have to pay attention to literally every possible sign that the software is not working. And I think the first step is actually just believing that when you type something or you build a feature, it's not going to work.”
Top AI engineers can intuit the quality curve of new features
“Really good AI people are able to intuit what the quality curve will look like for a new feature.”
Recruiting is an apprenticeship trade best learned from elite recruiters
“I think recruiting is one of those sort of trades that you need to learn by working as an apprentice with someone who's really good at recruiting.”
When a product fails to sell, the problem is almost always pipeline
“Like if your product isn't selling for some reason, then it's almost always pipeline.”
Starting a company without an idea prevents finding product-market fit
“And then I did something which I honestly don't think any founder should do. We're talking about product market fit. This is almost a surefire way to prevent yourself from finding product market fit. But I sat down and I said, okay, I want to start a company. …”
Raising venture capital does not validate idea quality or product-market fit
“Raising money was no indication of the quality of your idea or product market fit or chances for finding product market fit.”
Sales founders learn markets are uncontrollable earlier than engineering founders
“I think that sales or go-to-market founders often have this intuition earlier than engineering founders do is that markets and what people actually want are in many ways out of your control.”
Early founders must be ultra-paranoid about customer use-case repeatability
“I think in the world of early product building, you actually need to be like ultra paranoid about repeatability. And you need to assume that, you know, give yourself a little bit of credit and assume that you're somewhat charismatic or Entertaining or whatever…”
Impira reached millions in ARR without repeatable customer use cases
“We had a few million dollars in ARR, but very little repeatability across what they were doing. So when we tried to bring on a sales team and scale that what they were selling and their ability to scale the success from those early users into the end plus firs…”
Early AI teams preferred managed evals over brittle open-source tools
“We actually heard early on from people and allergic, no reaction to that. They were like, Hey, we were using open source stuff and it's super brittle and it breaks all the time. And we just evals suck and observability sucks and we just don't want to deal with…”
Braintrust built a terrible go-to-market motion to test product strength
“I wanted at Braintrust to, again, in the spirit of skepticism, Build a terrible go-to-market motion early on, but make the product, make Braintrust only successful if the product was so good that despite being grossly incompetent at selling and marketing our p…”
Intellectually fascinating technical problems efficiently drive peer-to-peer customer acquisition
“I think that people without us having to force them to or ask them to, they are just interested in solving the problem around evals and they find it intellectually fascinating on their own. And I think that leads to a lot of organic knowledge sharing. A lot of…”
Overhyped failure is much better for startups than dying in obscurity
“I think there's only one right answer, which is it's much better to be overhyped and then fail. I mean, obviously within the realms of morality and integrity, but I was, I sort of learned like, I don't want to be an obscure product that no one cares about.”
AI observability logs average 50KB per row compared to 900B traditionally
“Every span in brain trust land, which is like a row of something that you'd log in brain trust, the average size is
50 kilobytes.
In traditional observability, it's 900 bytes.”
Tantivi was the only search library to survive Braintrust's 100x benchmark
“There's one library called Tantivi, which is a Rust re-implementation of the Lucene search index, which is a very popular and well-regarded and used search technology. It's the technology behind Elasticsearch as well. And that was the only piece of code that w…”
Relational databases perform terribly at full-text search workloads
“Understandable, but staggering to me how bad relational databases are at full text search. Just, they're just awful.”
Braintrust deliberately excluded traditional enterprises early to focus on product engineers
“We say our product is for product engineering teams that are trying to incorporate AI into their core products and services. And that description excludes and excluded a lot of, ah, sort of traditional enterprises, for example, early on, and we were okay with …”
Silicon Valley multimodal AI workloads are video; enterprise workloads are PDFs
“So a lot of the cool Silicon Valley companies, their multimodal workloads are Mostly images and nowadays videos. A lot of the more traditional companies have a lot of PDFs.”
Braintrust operates with only one company-wide meeting per week
“Like, we don't have, we have one meeting per week as a company, and that's it.”
Goyal cuts off all meetings at noon to protect afternoon focus
“So I cut off meetings at noon every day. I don't take any meetings afternoon.”
Goyal has worked every weekend since joining the MemSQL founding team
“I work every weekend. And I have since I joined MemSQL.”
Requiring 40-to-50-person startup teams to work weekends is very challenging
“I think personally, my perspective is even if you're trying to hire like a pretty good, well-rounded and, you know, skilled group of people that's even, you know, 40 or 50 people, it's very challenging to get everyone to work on weekends. And so we don't, that…”
If a startup has to convince users, it lacks product-market fit
“I think if you have to convince people, it's super unlikely that you have product market fit. Whereas if without convincing people, they just find their way to the product and then they've, they're already convinced or they've convinced themselves very quickly…”
Alana Goyal and Elad Gil were the first investors in Braintrust
“Honestly, I think there's two people I'd point to and these were the first two people that invested in BrainTrust and really helped us. The first is Alana”
Pre-PMF startups must hire salespeople accustomed to working without PMF
“A classic example of this is when companies hire salespeople from extremely successful product-led companies, and they expect the salesperson to be good for their product-led company that does not have product market fit. That is not the right You actually wan…”