What Braintrust got right about product-market fit | Ankur Goyal (Founder and CEO)

Career Pivot Story

Ankur's journey into technology began with a dramatic shift during his senior year of high school in Pittsburgh. Originally planning to follow his parents' path as a doctor, he experienced what he calls a "midlife crisis in high school" when he took two pivotal classes.

The Turning Point:

1. AP Biology - Absolutely hated it, realizing medicine wasn't his calling
2. Linear Algebra - Completely fell in love with the mathematical concepts
3. PageRank Algorithm - Implemented it in Lisp, which was a transformative experience

The Decision Process:

• Applied to Carnegie Mellon University's Computer Science program
• Parents weren't initially supportive of the career change
• Started studying CS but continued searching for his true passion

Finding His Path:

• Interned at Microsoft and did academic research - hated both experiences
• Had his "second existential crisis" about traditional paths
• Made the bold decision to move to San Francisco and join MemSQL
• Dropped out of undergrad to work with Nikita and Eric at the startup
• Never looked back on the decision

Timestamp: [1:32-3:23]

Foundational Startup Lessons

From his first job out of college at MemSQL (now SingleStore), Ankur gained critical insights that shaped his approach to building companies.

Quality is Existential:

1. 95% to 99% Challenge - The craft of taking software from prototype to production-ready is extremely difficult
2. Paranoid Attention to Detail - Must pay attention to every broken test and customer issue
3. Real-World Stakes - A Goldman Sachs managing director told 23-year-old Ankur that if their software crashed, people could lose their jobs

Key Learning Areas:

Software Quality Mindset:

• Building a prototype is easy; building something that actually works is hard
• Senior engineer Adam from Microsoft exemplified the right approach - never brushing off issues
• Quality isn't just important, it's existential for the business

Recruiting Excellence:

• Recruiting really good engineers is not a transactional process
• Sometimes spent years recruiting a single person
• Required getting coffee quarterly and maintaining relationships
• Essential skill for any aspiring founder to learn

Customer Connection:

• At MemSQL, sold to technical users who shared his background
• Could relate intuitively to customer needs and problems
• This connection was crucial but completely taken for granted at the time

Timestamp: [3:38-7:02]

Quality-First Development Philosophy

Based on his experience at MemSQL, Ankur developed a systematic approach to ensuring software quality that goes beyond basic testing.

Core Mindset Shift:

Default Assumption: When you build a feature, assume it's not going to work

The Quality Framework:

Attention to Every Signal:

• Pay attention to literally every possible sign that software isn't working
• Don't dismiss broken tests or customer issues
• Treat each problem as a learning opportunity

Experience-Based Intuition:

1. Junior Engineers - Type something, test a few times, assume it works
2. Senior Engineers - Have strong intuition about potential failure points
3. Predictive Ability - Can anticipate quality issues based on feature complexity and scope

Modern Application:

• AI Development - Really good AI people can intuit what the quality curve will look like for new features
• Same principles apply across different technology domains
• Experience builds the ability to predict where problems will emerge

Practical Implementation:

• Question every assumption about functionality
• Build systematic processes for catching issues early
• Develop team culture where quality concerns are never brushed aside
• Invest in senior talent who can mentor others on quality mindset

Timestamp: [7:08-7:55]

Essential Insights:

1. Career Pivots Can Be Transformative - Ankur's shift from medicine to technology started with a single linear algebra class and implementing the PageRank algorithm
2. Quality is Existential - The difference between a prototype and production-ready software is enormous and requires paranoid attention to detail
3. Customer Connection Matters - Being able to relate to and understand your customers intuitively is crucial for product development

Actionable Insights:

• For Aspiring Founders: Spend time learning recruiting at companies that excel at it - it's one of the most challenging aspects of starting a company
• For Engineers: Develop the mindset that new features won't work by default and build systematic processes to catch issues
• For Product Development: Consider whether you can relate to your target customers - this connection significantly impacts your ability to build the right product

Timestamp: [1:09-7:55]

Product Quality Through Radical Attention to Detail

The Human Nature Challenge:

When users report issues at inconvenient times (like 6 PM on Friday), our natural tendency is to dismiss them:

• Button requiring double-clicks - Easy to rationalize as user error
• Intermittent 504 errors - Tempting to blame on user's setup or old browser
• One failure out of 20 attempts - Seems acceptable statistically

The Quality Mindset Shift:

Exceptional product builders take a radically different approach:

1. Razor focus on every detail - No issue is too small to investigate
2. Use complaints as validation - Each report confirms something is likely broken
3. Resist rationalization - Don't explain away problems with external factors

Why Product Quality Matters:

• Career-level stakes: Users bet their professional reputation on your software quality
• Scaling efficiency: High quality reduces support tickets and manual intervention
• Organizational expansion: Better quality drives adoption across more teams
• User investment: People invest precious time and energy based on product reliability

Timestamp: [8:02-10:05]

Advanced Recruiting Strategies for Startup Success

Learn Through Apprenticeship:

• Work with exceptional recruiters - Learn the trade through hands-on experience
• Challenge preconceived notions - Many intuitive approaches are counterproductive
• Avoid desperation tactics - Begging candidates creates poor company impressions

Time Investment Reality:

1. Accept the timeline - Finding 3-4 key people may take 1-2 years
2. Design around delays - Plan your company trajectory with realistic hiring expectations
3. Treat it like sales pipeline - Quality recruiting requires long-term investment

Rigorous Interview Standards:

• Challenge everyone equally - Even known quantities need thorough evaluation
• Separate relationship from assessment - Don't let familiarity cloud judgment
• Maintain high performance standards - Rigorous interviews protect team quality

Implementation Example:

At Braintrust, the founder and his brother spent 3-4 days crafting technical interview questions still used today, demonstrating the importance of systematic approach to evaluation.

Timestamp: [10:05-13:16]

The Journey from Employee to Entrepreneur

The Decision Framework:

After 5.5 years at MemSQL, Ankur used a simple but effective evaluation method:

• Regular self-assessment - Asked "Am I learning a lot?" every 6 months
• Honest recognition - When the answer shifted from "emphatically yes" to "no"
• Personal growth motivation - Wanted to push himself to do something harder

The Transition Strategy:

1. Strategic break period - Took 6 months off on Elad Gil's advice
2. Perfect timing insight - "Next time you might not be able to" (marriage, kids, responsibilities)
3. Life intersection - Met his future wife Elena on the last day of his break
4. No break the second time - Zero days off between previous job and Braintrust

The Founding Mistake:

Ankur admits to a critical error in approach:

• Technology-first thinking - Started with "I understand this technology, what company should I build?"
• Fundraising misconception - Didn't realize how easy it would be to raise money
• False validation - Mistook fundraising success for idea quality or product-market fit potential

The AI Timing Misjudgment:

5 years before ChatGPT, he believed:

• AI was ready to transform enterprise data usage
• Databases would evolve into AI-centric query tools
• This became the foundation for Impira (later acquired by Figma)

Timestamp: [13:16-15:55]

Essential Insights:

1. Product quality obsession - Taking every user complaint seriously, even minor ones, separates exceptional builders from average ones
2. Recruiting mastery requires apprenticeship - Learning from exceptional recruiters and accepting 1-2 year timelines for key hires
3. Founding motivations matter - Starting a company just for personal challenge often leads to product-market fit struggles

Actionable Insights:

• Implement rigorous interview processes even for known candidates to maintain team quality
• Design company plans around realistic hiring timelines rather than rushing recruitment
• Resist the urge to rationalize away user feedback - use complaints as validation that something needs fixing
• Don't mistake fundraising ease for idea validation or product-market fit potential

Timestamp: [8:02-15:55]

The Control Illusion in Technical Problem-Solving

Engineering founders often develop a dangerous bias from their technical background - the belief that if they can solve a technically challenging problem, people will automatically want it.

The Engineering Mindset:

• Complete Control: Technical problems feel entirely within your control - the difference between having a problem and solving it is literally typing keystrokes
• Linear Thinking: A 10,000-line PR might take just a few hours if you knew exactly what to type
• Solution-First Approach: Focus on what's technically challenging rather than what the market actually needs

The Market Reality:

• Markets are uncontrollable: What people actually want and when they want it is largely outside your control
• Timing matters: Understanding when people need something is as crucial as building it
• Demand isn't guaranteed: Technical complexity doesn't equal market demand

Key Learning:

Sales and go-to-market founders often develop market intuition earlier than engineering founders because they're constantly interfacing with customer needs and market dynamics.

Timestamp: [16:02-17:45]

The Charismatic Founder Trap

Impira's first year demonstrated a common startup pitfall where founder charisma masks fundamental product-market fit issues.

The Initial Success:

• Built an amorphous tool for querying images, videos, and documents with AI
• Got people using it and paying for it
• Achieved a few million dollars in ARR

The Hidden Problem:

• Interest vs. Need: Customers were primarily paying for interest in the founders and help solving problems
• Zero Repeatability: Very little consistency across what different customers were asking for
• Custom Solutions: Each customer required product modifications for their specific needs

The Scaling Challenge:

When they tried to bring on a sales team and scale:

• No Replicable Process: What early customers bought couldn't be easily sold to new customers
• N+1 Problem: Success with early users didn't translate to the next user
• Sales Team Struggles: New salespeople couldn't replicate the founder's success

Critical Insight:

Founders need to be "ultra paranoid about repeatability" and assume that customer interest might be for the wrong reasons, even when people seem genuinely interested in the product.

Timestamp: [17:51-19:33]

The Most Dangerous Startup Trap

Weak product-market fit represents one of the most insidious challenges for startups because it creates the illusion of progress while lacking sustainable growth foundations.

Characteristics of Weak Product-Market Fit:

• Some customers: You have paying customers, so the world isn't telling you you're completely wrong
• Revenue generation: Money is coming in, creating false confidence
• Limited repeatability: Success with one customer doesn't easily translate to others
• Custom solutions: Each customer requires significant customization

Why It's More Dangerous Than No Customers:

• False signals: Having some customers makes you think you're on the right track
• Time trap: You can spend years pushing down this path instead of pivoting
• Resource drain: Constantly customizing for different customers without building scalable solutions
• Delayed recognition: The problem becomes apparent only when trying to scale

The Alternative Reality:

Having no customers at all would actually be preferable because:

• Clear feedback: The market is clearly telling you something is wrong
• Faster iteration: You're forced to pivot or find real product-market fit quickly
• No false hope: You can't fool yourself into thinking you're succeeding

Timestamp: [19:33-20:08]

From Repeated Pain Point to Startup Opportunity

Braintrust emerged from Ankur experiencing the exact same technical challenge at two different companies, revealing a broader market need.

The Recurring Problem - Evals:

At Impira:

• Customers using the platform for invoices and bank statements
• Model or prompt updates would improve invoices but make bank statements worse
• Financial customers couldn't tolerate regression in accuracy
• Had to build internal tooling to measure changes during development

At Figma:

• Leading the AI team after the Impira acquisition
• Encountered exactly the same evaluation challenges
• Built the same internal tooling again
• Foundation for Figma's current eval approach was set during this period

The Realization Moment:

During regular conversations with El (a recruiter), they discussed the parallels:

• Pattern Recognition: Same tooling built twice at different companies
• Market Timing: When Ankur was at Impira, only ~3 companies were working on AI; now many companies care about it
• Logical Conclusion: If two companies needed the same tooling, others probably would too

The Validation Approach:

• Initially skeptical about the idea as a business opportunity
• Decided to incubate as a side project first
• Created conditions that would need to be true for it to be viable
• Systematic approach to testing the hypothesis before committing

Timestamp: [20:08-22:03]

Systematic Market Validation Strategy

Braintrust used a methodical approach to validate their hypothesis by testing with industry leaders before committing to the business.

The Target List Strategy:

• 50 companies: Identified companies ahead of the curve in AI development
• Key players: Notion, Airtable, Zapier, Coda, Instacart - many became first customers
• Complete mapping: Listed company name, AI team leader, and founders for each target
• Network leverage: Used El's incredible network and Ankur's connections to get meetings

The Interview Approach:

• Open-ended questions: Designed non-leading questions about biggest AI challenges
• Hypothesis testing: Specifically assessing whether the eval problem was widespread
• Antagonistic mindset: Approached with skepticism, almost wanting to disprove the idea
• Systematic documentation: Carefully recorded responses and patterns

Surprising Market Learning - Open Source Rejection:

Initial Assumption: Coming from MemSQL's closed-source background, assumed people wanted open-source solutions

Market Reality:

• Allergic reaction to open-source suggestion
• Pain with existing tools: "We're using open source stuff and it's super brittle"
• Reliability concerns: "It breaks all the time and evals suck and observability sucks"
• Clear preference: "We just want something we can install and it works"

Validation Outcome:

This systematic approach revealed both the market need and crucial insights about customer preferences that shaped Braintrust's product strategy.

Timestamp: [22:03-23:54]

Essential Insights:

1. Engineering founder bias - Technical problem-solving ability creates dangerous assumption that market demand follows technical capability
2. Weak product-market fit trap - Having some paying customers with no repeatability is more dangerous than having no customers at all
3. Systematic validation approach - Testing market hypotheses with 50 target companies revealed both demand and crucial product preferences

Actionable Insights:

• Be "ultra paranoid about repeatability" when early customers show interest - assume they might be interested for the wrong reasons
• Write down specific conditions that must be true for your business idea to work, then systematically test them
• Challenge your own biases through customer interviews - Braintrust discovered customers wanted the opposite of what founders assumed about open source

Timestamp: [16:02-23:54]

Early Customer Validation Through Direct Feedback

The Zapier Breakthrough:

1. Immediate Problem Recognition - During user interviews, everyone consistently identified evals as their main AI development challenge
2. Proactive Customer Engagement - Zapier's CTO Brian sent an unsolicited email asking if Braintrust was actually fixing the eval problem because it was becoming serious for them
3. Direct Collaboration Channel - Zapier created a shared Slack channel to directly tell the team what to build

Rapid Prototype Success:

• 5-Day Timeline - Built a terrible, ugly prototype that barely ran
• Immediate Adoption - Zapier team started using it despite its poor quality
• Visual Feedback Loop - Engineers took screenshots of the UI and annotated them with feature requests

Early Adopter Pattern:

• KOD Validation - David (now at Figma) became another early user with similar needs
• Consistent Requests - Both companies asked for the same solutions, particularly a prompt playground
• Universal Adoption - Once the prompt playground was built, it became the most requested feature until shipped

Timestamp: [24:00-25:43]

Family Partnership in Tech Entrepreneurship

The Switzerland Wedding Realization:

• Limited Quality Time - Brothers realized they don't spend enough time together despite wanting to
• Shared Passion - When they do spend time together, they naturally talk about computer science
• Social Friction - Their technical conversations irritate family members and friends during social gatherings

Strategic Partnership Decision:

1. Manu's Experience - His brother had worked at Neuro for six years, bringing substantial expertise
2. Natural Synergy - Both enjoyed discussing computer science and technical challenges
3. Time Optimization - Working together would allow them to spend more time together while pursuing their shared interests

The Timing Alignment:

• Ankur's Availability - He wasn't actively looking to start a company but recognized the rare opportunity
• Enjoyment Factor - He genuinely enjoyed spending time with Elad (his co-founder)
• Perfect Convergence - The technical opportunity, family partnership, and co-founder chemistry aligned simultaneously

This family-business combination became the foundational kernel of Braintrust's founding team structure.

Timestamp: [25:48-26:43]

Balancing Paranoia with Strategic Optimism

The Founder Dichotomy:

• Universal Pattern - Every good founder and investor exhibits strange dichotomy between being paranoid/skeptical and hugely optimistic
• Evolved Approach - Ankur's formula has developed over time through experience, though he hasn't perfected it
• Strategic Application - Different mindsets applied to different aspects of business building

Skepticism Framework:

Things to Be Skeptical About (Outside Your Control):

1. Market Dynamics - Whether the market would actually adopt eval solutions
2. Customer Behavior - Whether people would think about evals the way he expected
3. Personal Readiness - His own capabilities after a stressful acquisition experience

Things to Be Optimistic About (Within Your Control):

1. Technical Expertise - Building software that involves crunching data
2. Network Quality - The value of his and Elad's connections in the tech community
3. Execution Capability - Despite self-doubt about being a "magnificent CEO"

Post-Acquisition Mindset:

• Stress and Fatigue - Coming off an acquisition that "wasn't the best in the world"
• Leadership Uncertainty - Even considered hiring a CEO or finding a co-founder CEO
• Elad's Encouragement - Co-founder pushed him to lead, citing his previous experience with startup challenges

Timestamp: [26:49-29:17]

Product-First Growth Through Strategic Incompetence

Learning from Previous Companies:

• Over-Sophisticated Sales - At both Impira and MemSQL, go-to-market was way more sophisticated than the product
• Sales-Heavy Approach - Great sales talent, processes, website metrics, and aggressive opportunity management
• Misaligned Priorities - Strong sales capabilities masked product weaknesses

The Intentional Handicap Strategy:

1. Terrible Go-to-Market Motion - Deliberately built incompetent selling and marketing processes
2. Product-Only Success Metric - Made Braintrust successful only if the product was so good it would take off despite poor sales
3. No Pricing Discussions - Didn't talk to early users about pricing at all

Organic Customer Conversion:

• Production-Driven Pricing - Only discussed pricing when customers put the product in production
• Customer-Initiated Commerce - Customers asked to pay due to liability concerns: "Can we please pay you?"
• Timeline Success - Started in August, first three customers requested to pay by November

Unexpected Results:

Organic Growth Phenomena:

• Random Household Names - Major companies discovered and started using Braintrust organically
• Word-of-Mouth Spread - Customers found the product through friends or Twitter mentions
• Exceeded All Targets - Beat every sales target throughout the year with purely inbound traffic

Simplified Sales Process:

• Generic Demos - Unlike previous companies where he'd spend days preparing customer-specific demos
• Universal Use Cases - Could show the same generic use case to everyone because all customers wanted the same thing
• Volume Management - Too many customer meetings to customize each one individually

Timestamp: [29:23-31:54]

Essential Insights:

1. Customer-Driven Product Development - Zapier's proactive engagement and direct feedback loop validated Braintrust's core value proposition within days of building a terrible prototype
2. Strategic Skepticism Framework - Successful founders balance paranoia about external factors (market, customer behavior) with optimism about controllable elements (technical skills, network, execution)
3. Intentional Go-to-Market Weakness - Building deliberately poor sales processes forces product excellence and creates sustainable organic growth patterns

Actionable Insights:

• Create direct feedback channels with early customers who can tell you exactly what to build
• Apply skepticism to market assumptions while maintaining confidence in your core competencies
• Consider handicapping your sales process early to ensure product-market fit isn't masked by strong sales capabilities
• Allow customers to discover pricing needs organically through production usage rather than premature monetization discussions

Timestamp: [24:00-31:54]

Immediate Customer Recognition and Demand

Key Indicators of Strong Product-Market Fit:

1. Instant Visual Recognition - Customers literally point at the screen saying "I need exactly that now"
2. Zero Convincing Required - No effort needed to persuade customers they need the product
3. Organic Demand Generation - Customers come to you rather than requiring sales outreach

The Braintrust Experience:

• Immediate Response: People would look at demos and immediately recognize the solution to their problem
• Effortless Sales Process: Complete contrast to previous companies where convincing customers was necessary
• Clear Market Signal: When you spend almost no energy trying to convince anyone they need your product

Comparison to Previous Ventures:

The CEO notes this was a "very very different experience" compared to Impira and MemSQL, where significant effort was required to demonstrate product necessity.

Timestamp: [32:00-32:22]

Problem-Focused Community Engagement

Essential Market Characteristics:

1. Large, Urgent, and Repeatable - Problems that affect multiple customers consistently
2. Intellectual Fascination - Customers find the problem space genuinely interesting
3. Organic Knowledge Sharing - Community naturally discusses solutions without prompting

The DBT Parallel:

• Community Formation: Data engineers naturally sought each other out to discuss data warehouse challenges
• Problem-First Approach: Focus was on the underlying problem, not specific tool features
• Valuable Outcomes: Solving data wrangling created hugely valuable business results

Braintrust's Market Advantage:

Natural Community Dynamics:

• Customers enjoy discussing evaluation (eval) problems intellectually
• Organic knowledge sharing occurs without company intervention
• Success with one customer efficiently translates to others

Content-Driven Growth:

• Blog posts about customer implementations (like Notion) drive interest
• Readers learn about evaluation approaches in March, then contact in July
• Educational content creates delayed but qualified leads

Timestamp: [32:22-34:24]

Choosing Visibility Over Perfection

The Launch Strategy Decision:

• Timeline: Started building in August, launched publicly in September
• Media Coverage: TechCrunch coverage and social media buzz from friends
• Early Validation: High-taste engineers at companies like Zapier were already using the product

Key Philosophy - Two Failure Modes:

1. Die in Obscurity - Build in stealth mode too long and never gain traction
2. Get Overhyped and Fail - Launch early with imperfect product but gain visibility

The Right Choice:

Better to be overhyped and fail (within moral and integrity bounds) than to remain an obscure product nobody cares about.

Learning from Impira:

• Previous Mistake: Protected company name and stayed stealth too long
• Correction Applied: Put Braintrust name out immediately to create buzz
• Validation Threshold: If valuable to high-taste early users, it will be valuable to others

Product State at Launch:

• Honest Assessment: "Our product sucks. It can improve. The UI was terrible for a long time."
• Value Delivery: Despite imperfections, it was providing clear value to users
• Continuous Improvement: Focus on making it better day after day

Timestamp: [34:30-36:22]

Customer-Driven Development Over Technical Moats

First Six Months Approach:

• Minimal Technology: Built very little interesting technology initially
• Customer Focus: Prioritized user needs over technical sophistication
• VC Concerns: Investors questioned durability and competitive moats

Ignoring Technical Concerns:

Common VC Questions:

• What durable technology are you building?
• What's your moat against competitors?
• How will you prevent customers from building this internally?

The Response:

"I just didn't care because I was talking to customers and they didn't care"

The Notion Catalyst:

Advanced Use Case Discovery:

• Leading Customer: Notion was far ahead in AI product adoption
• Unique Needs: Searching through logs in ways traditional observability couldn't handle
• Scale Challenge: Growing rapidly and pushing current system limits

Technical Reality Check:

• Data Size Difference: AI spans average 50KB vs traditional observability 900 bytes
• Search Requirements: Full-text search through enormous rows with prompts and text
• Infrastructure Limitations: Existing database technologies couldn't solve the problem

The Benchmark Solution:

1. Stress Testing: Built benchmarks 100x more demanding than Notion's current needs
2. Technology Evaluation: Tested PostgreSQL, ClickHouse, Snowflake, Redshift, and others
3. Discovery: Only Tantivy (Rust re-implementation of Lucene) could handle the requirements
4. Revelation: Traditional relational databases are "just awful" at full-text search

Timestamp: [36:42-39:57]

Essential Insights:

1. Product-Market Fit Signals - When customers immediately recognize and demand your solution without convincing, you've found strong product-market fit
2. Market Selection Criteria - Good markets have problems that are large, urgent, repeatable, and intellectually fascinating to customers
3. Launch Timing Philosophy - Better to launch early with an imperfect product and risk being overhyped than to die in obscurity

Actionable Insights:

• Focus on customer validation over technical perfection in early stages
• Build communities around problems, not just solutions, for organic growth
• Use leading customers' advanced needs to drive technical innovation
• Prioritize customer feedback over investor concerns about competitive moats
• Launch publicly as soon as you have evidence of value delivery to quality users

Timestamp: [32:00-39:57]

Building Brainstorm: From Customer Need to Technical Solution

The Genesis:

1. Customer-Driven Development - Notion pushed Braintrust to build novel technology after a year of focusing purely on customer needs
2. Strategic Pivot - The team abandoned their "no moats" philosophy when they realized existing solutions couldn't solve the problem
3. Rapid Execution - Three experienced engineers built and shipped a purpose-built database in record time

The Team and Timeline:

• October: Broke ground with three engineers (Ankur, Manu, and Austin)
• January: Shipped to Notion for testing and feedback
• February: Released to all customers

Why It Worked So Well:

• Deep Expertise: All three engineers had extensive experience with this type of technology
• Precise Understanding: A full year of suffering through the problem gave them exact requirements
• Purpose-Built Focus: Brainstorm is extremely specialized - won't work as a data warehouse replacement but excels at Brain Trust's specific use case
• Strong Relationships: Team members had known each other for years (Ankur and Manu for 30 years, Austin for a decade)

Key Success Factors:

• No margin for error but extreme confidence in requirements
• Customer partnership with Notion for early testing and issue identification
• Accumulated experience from trying to solve the problem without proper tooling

Timestamp: [40:04-42:06]

Strategic Framework for AI Product Development

Long-Term vs Short-Term Betting Strategy:

1. Historical Experience - Building solutions to problems they experienced at Impira that people still face today
2. Explicit Classification - Clearly categorizing which features are long-term investments vs short-term adaptations
3. No Rejection of Short-Term - Embracing temporary solutions while maintaining strategic focus

Obvious Long-Term Bets:

• Data Volume Growth: People will log more AI-related data over time
• Prompt Size Expansion: Prompts will continue getting bigger and more complex
• Infrastructure Investment: Technology for easy, scalable, and cheap data handling will remain relevant

Example of Strategic Evolution - The Playground:

Initial Resistance:

• Ankur preferred traditional development tools (Cursor, IDE, Vim, source control)
• Viewed playground as potentially inferior to professional development environments

Market Reality:

• Diverse User Base: Product people, support teams, even doctors writing prompts
• Developer Adoption: Even engineers embraced the playground for faster AI iteration
• Professional Demands: Users wanted full IDE functionality within the playground

Strategic Response:

• 8 months ago: Made explicit decision to treat playground as a real IDE
• Quality Bar: Engineering with professional IDE standards in mind
• Recent Investment: Months spent re-engineering React state to eliminate technical debt
• Features Added: Stateful operation, Figma-style multiplayer collaboration, eval integration

Timestamp: [42:12-45:40]

Dynamic Feature Management in Fast-Moving AI Space

Constant Change Examples:

• Terminology Evolution: "Agent" means something different today than yesterday, will change again tomorrow
• User Expectations: People want Braintrust to be accessible for exploring current AI trends
• Feature Relevance: What matters to users shifts rapidly with AI developments

Adaptive Strategies:

Flexible Content Delivery:

1. Cookbooks: Constantly updated tutorials for trying new AI approaches
2. UI Iteration: Regular updates to what's highlighted vs. what's collected in background
3. Feature Prominence: Adjusting what's front-and-center based on current user needs

Real-World Example - Cache Token Tracking:

• Background Collection: Always tracked cache token counts
• Emerging Relevance: Pricing implications became topical issue for users
• Strategic Response: Made cache token tracking and pricing analysis very easy and prominent
• Temporal Awareness: May not be important in 6 months if models change dramatically

Future-Proofing Approach:

Hypothetical Scenario - Tool Background Processing:

• Emerging Pattern: OpenAI introducing tools running in model background
• Potential Impact: New pricing considerations if this becomes widespread
• Adaptive Response: Would prioritize this on monitoring page if it becomes user priority
• Bucket Strategy: Different engineering durability for different types of features

Core Philosophy:

• Different Buckets: Separate treatment for foundational vs. trend-responsive features
• User-Centric: Following where user attention and needs are focused
• Engineering Flexibility: Varying levels of robustness based on feature longevity expectations

Timestamp: [45:40-47:56]

Essential Insights:

1. Customer-Driven Innovation - Notion's specific needs pushed Braintrust to build Brainstorm, a custom database delivered in just 2.5 months through deep expertise and precise requirements understanding
2. Strategic Betting Framework - Braintrust explicitly categorizes features as long-term vs short-term bets, investing in obvious trends like data volume growth while staying adaptable to rapid AI changes
3. Professional Tool Evolution - The playground evolved from a simple testing tool to a full IDE replacement based on diverse user needs, requiring months of re-engineering for professional quality standards

Actionable Insights:

• Build purpose-built solutions rather than trying to be everything to everyone - Brainstorm excels at specific use cases but won't work as a general data warehouse
• Embrace both foundational investments and trend-responsive features with different engineering approaches and durability expectations
• Listen to unexpected user demands even when they conflict with your preferences - the playground's success came from accepting non-developer users' needs

Timestamp: [40:04-47:56]

Strategic Customer Focus and Market Positioning

Braintrust maintained laser focus by being extremely clear about their target market and sticking to it, even when it meant excluding potential customers.

Core Product Definition:

• Target Audience: Product engineering teams incorporating AI into their core products and services
• Deliberate Exclusions: Traditional enterprises that weren't ready for this approach
• Strategic Patience: Maintained focus while staying in touch with excluded prospects

Market Evolution Strategy:

1. Early Hypothesis: Believed that advanced AI engineering practices would eventually spread to traditional enterprises
2. Validation Over Time: Many initially excluded enterprises eventually came around as they began building AI into external apps and internal tools
3. Workflow Conviction: Remained firm about what they believed was the right workflow for AI engineering

Adaptation Without Compromise:

• Infrastructure Adjustments: Added Azure support for enterprise security requirements
• Content Format Evolution: Enhanced PDF support as traditional companies had different multimodal needs than Silicon Valley companies (images/videos vs. PDFs)
• Core Product Integrity: These adaptations felt comfortable within their repeatable product framework

The key was distinguishing between surface-level adaptations (deployment, file formats) versus fundamental product pivots (custom LLM training, fine-tuning focus) that would have pulled them away from their core vision.

Timestamp: [48:01-50:42]

Building Authentic Partnerships in the AI Ecosystem

Braintrust's close association with leading AI companies stems from genuine relationships built through shared challenges and mutual support, rather than manufactured marketing efforts.

Foundation of Trust:

• Shared Struggles: Companies have been through difficult technical challenges together
• Mutual Product Testing: Braintrust uses alpha versions of partner companies' AI products while partners break and test Braintrust's platform
• Long-term Relationships: Many relationships predate Braintrust's founding
• Personal Connections: Regular social interactions including lunches and dinners

Natural Advocacy Approach:

1. Authentic Enthusiasm: Partners genuinely celebrate each other's successes, similar to how people feel pride when colleagues or new hires excel
2. Problem-Focused Collaboration: United by shared technical challenges in AI engineering
3. Organic Promotion: Success stories emerge naturally from real product value rather than coordinated marketing

Leadership Philosophy:

• Customer-First Mentality: Focus on delivering exceptional products to existing customers
• Beyond Transactions: Care about customers' project success and career growth beyond just product usage
• Authentic Approach: Prefer genuine relationship building over manufactured promotional strategies
• Long-term Perspective: Belief that exceptional product delivery and authentic care will translate to commercial success

Timestamp: [50:48-53:36]

Strategic Bet Clarity and Conviction Management

Braintrust's success stems from being extremely clear about which bets to take with conviction versus which areas to remain scientifically skeptical about.

Conviction Bets:

• Customer Focus: Bet heavily on listening exactly to what their specific target customers needed
• Market Pattern: Believed that advanced AI engineering practices would eventually spread beyond cutting-edge companies
• Core Strengths: Confident in their data processing capabilities and UI quality

Areas of Strategic Skepticism:

• Market Validation: Initially very skeptical about whether the company should exist at all
• Technical Decisions: Questioned whether to build proprietary database technology
• Market Risks: Acknowledged risks around AI hype cycles and whether their customer patterns would prove universal

Benefits of Clear Bet Framework:

1. Internal Focus: Helps concentrate thinking and decision-making energy
2. Recruitment Clarity: Enables transparent conversations with candidates about company risks and opportunities
3. Honest Risk Assessment: Can clearly articulate what they believe they'll win on versus genuine uncertainties
4. Candidate Alignment: Allows potential hires to make informed decisions about whether they believe in the company's specific bets

Practical Application:

When recruiting, Ankur can tell candidates exactly where Braintrust expects to win (data processing, UI quality) versus the real risks (AI market sustainability, customer pattern validity, major industry changes). This transparency helps attract people who genuinely believe in their specific vision rather than those who need to be convinced.

Timestamp: [54:01-55:53]

Essential Insights:

1. Strategic Focus Pays Off - Braintrust succeeded by being extremely specific about their target market (product engineering teams building AI into core products) and maintaining that focus even when excluding potential customers
2. Authentic Relationships Drive Growth - Their strong position in the AI ecosystem comes from genuine partnerships built through shared technical challenges, mutual product testing, and long-term personal relationships rather than manufactured marketing
3. Bet Clarity Enables Conviction - Success requires being explicit about which areas deserve full conviction versus which warrant scientific skepticism, helping both internal decision-making and transparent recruitment

Actionable Insights:

• Define your ideal customer profile precisely and stick to it, even if it means saying no to revenue opportunities that don't fit
• Build genuine relationships in your industry through shared problem-solving and mutual support rather than transactional networking
• Clearly articulate your company's core bets and risk areas to focus internal thinking and attract aligned team members
• Adapt surface-level features for market needs while maintaining core product integrity and vision

Timestamp: [48:01-55:53]

Hiring and Cultural Alignment Challenges

Key Hiring Pitfalls:

1. Firing too late - The most common mistake that compounds other issues
2. Creating departmental factions - When early hires in different departments conflict with each other
3. Missing the real problem person - Sometimes you fire the wrong person while the actual source of drama remains

The Faction Problem:

• Cross-functional conflicts: Early hires from different departments often clash
• Leadership responsibility: These conflicts are largely the founder's fault for not managing properly
• Cascading effects: One underperforming person can drag down others who were previously productive
• The ablation test: After removing someone, you realize another person was actually the root cause

Cultural Misfits vs. Performance Issues:

• Scaling myths: The idea that "this person was good early but not later" may not always be accurate
• Drama contributors: Some people create ongoing conflicts that affect team dynamics
• Performance correlation: People who create interpersonal issues often also underperform eventually

Timestamp: [56:00-1:03:27]

Intentional Culture Design vs. Regression to the Mean

Braintrust's Specific Cultural Choices:

• One meeting per week - Company-wide meeting limit to maximize focus time
• No afternoon meetings - CEO cuts off meetings at noon for deep work
• Clear work style preferences: Perfect fit for designers who code, product managers who prototype, and engineers who care about product problems
• Not for everyone: Explicitly not suitable for people who need frequent meetings to accomplish work

Learning from Previous Experience:

• Impira's mistake: Built a culture that "regressed to the mean" by trying to accommodate everyone
• The empathy trap: First-time founders with empathy tend to listen to too many requests that aren't critical to business success
• Specificity over universality: Better to be a great place for some people than mediocre for everyone

Implementation Philosophy:

1. Personal preferences as seeds - Start with founder's clear perspective on how work gets done
2. Attract aligned people - Clear culture attracts those who share fundamental values
3. Trial and error approach - Honest admission that culture design involves learning through mistakes

Timestamp: [56:23-58:13]

The Challenge of Personal vs. Company Values

Personal Traits That Became Company Culture:

• No afternoon meetings: CEO's personal need for focused work time became company policy
• Minimal meeting culture: Recognition that work doesn't happen in meetings for many people
• Independence and autonomy: Core values that attract like-minded team members

Personal Traits That Stayed Personal:

• Working every weekend: CEO works weekends since joining MemSQL but doesn't require this from team
• Individual work styles: Recognizing people recharge and function differently
• Implementation details: Distinguishing between core values and specific work habits

The Decision Framework:

1. Trial and error approach - Honest admission of learning through mistakes over time
2. Competence over conformity - Found many highly competent people who don't share all personal work habits
3. Values vs. implementation - Separating fundamental values (hard work, independence) from specific behaviors (weekend work)
4. Long-term learning - Evolution from early MemSQL days where weekend work might have been required

The Hiring Reality:

• Mishiring and misfiring - Acknowledging a long history of learning through mistakes
• Scale considerations - Difficulty getting 40-50 people to all work weekends, even if some companies do it successfully
• Talent pool impact - Overly rigid requirements can limit access to skilled, well-rounded people

Timestamp: [58:18-1:01:35]

Beyond Metrics: The Unstoppable Usage Signal

The Ultimate Product-Market Fit Indicator:

• Can't stop people from using it: Users continue despite infrastructure problems
• Annoying level of demand: CEO finds it "almost annoying" when infrastructure is "on fire" but users won't stop
• Organic, persistent usage: People use the product regardless of technical issues

Philosophical Perspective on PMF:

• "When you feel it, you know it": Echoing the classic startup wisdom about product-market fit
• Experience-dependent understanding: If you haven't felt it, you can't understand what it means
• Intuitive recognition: True PMF becomes obvious through user behavior rather than metrics alone

Practical Manifestation:

• Infrastructure strain: High usage that pushes technical limits
• User persistence: Continued engagement despite technical problems
• Founder frustration paradox: Being annoyed by success because of operational challenges

Timestamp: [1:03:27-1:03:59]

Essential Insights:

1. Cultural specificity beats universality - Better to create a great workplace for some people than a mediocre one for everyone
2. Hiring mistakes compound - Firing too late and missing the real source of team conflicts are common founder pitfalls
3. Personal traits require careful curation - Not all founder preferences should become company culture; distinguish between core values and implementation details

Actionable Insights:

• Design company culture intentionally rather than letting it "regress to the mean" by accommodating everyone
• Use work trials to ensure cultural alignment before hiring, showing candidates real product issues upfront
• Separate fundamental values (hard work, autonomy) from personal work habits (weekend work, specific schedules)
• Recognize true product-market fit when users won't stop using your product despite technical problems
• Accept that culture design involves trial and error, with honest acknowledgment of learning through mistakes

Timestamp: [56:00-1:03:59]

Recognizing True Product-Market Fit

The Magnetic Pull Phenomenon:

1. Overwhelming Usage Demand - Users constantly push the limits of your product, even to the point where they strain your infrastructure
2. Effortless User Acquisition - People find their way to the product without convincing and quickly convince themselves of its value
3. Unstoppable User Behavior - Both existing and new users are magnetically pulled into using the product constantly

Key Warning Signs to Avoid:

• Having to convince people - If you need to persuade users to adopt your product, it's super unlikely you have product-market fit
• Resistance to usage - Users who need constant encouragement or incentives to engage

The Infrastructure Stress Test:

When users are so engaged that they're asking you to reduce their ability to use the product (like running fewer experiments), that's actually a very good thing. It indicates people can't get enough of what you're building.

The Self-Selection Principle:

True product-market fit means users are already convinced or convince themselves very quickly - there's no sales process required for adoption.

Timestamp: [1:04:06-1:05:01]

Key Mentors in Early Company Building

Alena - The People-Focused Investor:

• Background: Learned sales from her father, then transitioned to product work
• Unique Combination: Highly technical yet deeply understands customer and talent acquisition
• Communication Style: Easy for technical founders to relate to without feeling "too salesy"
• Core Strength: Exceptional taste in identifying the right customers and candidates for company success

The Meta Lesson for Technical Founders:

Vulnerability and Surrender - Technical founders need to make themselves vulnerable to people wired differently around:

• People dynamics
• Market understanding
• Customer relationships

This collaboration is essential because technical skills alone don't naturally translate to understanding what makes companies successful with people and markets.

Why This Partnership Matters:

For "super nerdy" founders, working with people-oriented advisors feels unnatural but is crucial for building companies that succeed beyond just having great products.

Timestamp: [1:05:18-1:06:48]

The "Because Of vs Despite" Framework

Core Principle:

Companies succeed because of certain things and despite other things. This means you shouldn't emulate everything successful companies do.

The Sales Hiring Paradox:

Wrong Approach: Hiring salespeople from extremely successful product-led companies Why It Fails: These salespeople never had to get good at selling because their previous company had strong product-market fit

The Right Approach:

Hire from struggle: Look for salespeople who worked at companies without product-market fit Why It Works: They learned to sell in "hard mode" and can transition to "easy mode" when you achieve product-market fit

Real Examples:

1. Brian (Braintrust's sales leader): Worked at a company with little product-market fit, then joined Grafana
2. Ron from Databricks: Similar path from low product-market fit company to Databricks success

The Counterintuitive Truth:

The transition from "something very hard to sell to easy mode" creates better sales leaders than hiring from companies that never had to develop those hard-earned selling skills.

Timestamp: [1:07:12-1:08:37]

The Trust-Building Process with Advisors

The Pattern Recognition:

After knowing Elad for 7 years, Ankur noticed a consistent pattern:

• Every time Elad believed in something that didn't make intuitive sense
• Every time Ankur didn't listen to Elad's advice
• Literally every time, Ankur was wrong

The Braintrust Market Example:

Ankur's Initial Skepticism: Very doubtful about the AI/ML market opportunity Reasoning: Didn't want to admit the market existed, plus concerns about CI/CD and other technical factors Elad's Position: Really bullish about the market potential

The Rewiring Process:

New Mental Framework: "If Elad is pushing on something and I think it doesn't make sense intuitively, then Elad is probably right"

Key Requirements for This Dynamic:

1. Time Investment: Building rapport over years (7+ years in this case)
2. Trust Development: Reaching a point where you can suspend disbelief
3. Relationship Building: Creating a foundation where counterintuitive advice can be received

The Overnight Myth:

This type of advisor relationship doesn't happen overnight - it requires sustained interaction and proven track record of accurate guidance.

Timestamp: [1:09:02-1:09:46]

Essential Insights:

1. Product-Market Fit Signals - True PMF shows as magnetic user pull and infrastructure strain from overwhelming usage, not from convincing people to adopt
2. Mentor Complementarity - Technical founders need people-focused advisors who understand markets and customer dynamics differently than engineers
3. Counterintuitive Hiring - The best salespeople often come from companies without product-market fit, where they learned to sell in hard mode

Actionable Insights:

• Look for users who can't help themselves from using your product constantly as a PMF indicator
• Make yourself vulnerable to advisors wired around people and markets, not just technical excellence
• When hiring sales talent, consider candidates who succeeded despite difficult selling environments rather than those who only knew easy mode
• Build long-term relationships with experienced advisors and learn to trust their counterintuitive insights over time

Timestamp: [1:04:06-1:09:52]