How to Build a Product that Hits PMF on Day 1 | Granola, Christopher Pedregal
These notes were generated by AI and may contain inaccuracies.
Talking to users is the lifeblood of building product. Users often misunderstand products. Example: YouTube video of product designer user test where user correctly places square in square hole, but tries circle in square hole, showing complete misunderstanding. Feedback cycles must be short; if it takes a month, it's almost not worth doing because original thinking is forgotten. Do not just do whatever users ask—users say many things, sometimes contradict, and what they think they want differs from what they need. Prioritized list of user needs does not work well.
Chris Pedregal, co-founder and CEO of Granola, an AI notepad for meetings like Apple Notes that listens in background, fleshes out raw notes into great notes at meeting end. Launched 1.5 years ago with 4 people, now 35 people, raised Series B at $250 million valuation, growing quickly. Loves building tools that improve lives. Born in '86, Google launched in '98 when he was 12, fascinated in small town upstate New York; internet was window to world, idolized Silicon Valley. Studied computer science in university to understand computers, internet, building for internet.
In security class, learned cryptography is magic but most vulnerabilities from people or poor system design—human interaction crucial. Discovered human-computer interaction field: study of how people use technology and designing for them. To build fantastic product, deeply understand user.
Products should feel like they have soul: coherent, consistent, from same values. Easy to tell, hard to build as complexity and people increase, leading to differing values and inconsistent screens. Famous principle: product design reveals company organizational structure—bad. Analogy: best friend predictable in reactions based on values; if acts like principal or boss, disorienting. Product with soul comes from deep values and personality, not company structure.
Philosophy: explore then exploit. Did not launch Granola for a year, allowing drastic changes. First 9-10 months added features, screens. Then cut 50% to focus on most important. Hard post-launch with users. Explore phase: figure solution shape, try many things. Exploit phase: polish the right shape—details matter, but wrong shape wastes effort. Hard to know right shape; launched to learn faster from more users, not realizing product-market fit (PMF) on day one. Took 6 months to recognize. Onboarded 2-3 people daily pre-launch for feedback; post-launch heard same things repeatedly.
Facebook team worked 6-9 months on Dropbox-like music file sharing post-launch, unsure of PMF despite being generational company. Mistake: assuming known pain point and solution, focusing on execution. Metaphor: video game map starts gray, reveals as you explore—terrain unknown until probed. Cannot design perfect thing in isolation; put stuff out, get feedback. Systematize user talks; lower activation energy company-wide.
2013: Started edtech Socratic, AI tutor for high school kids in NYC, ran 5 years, acquired by Google. Hard to talk to students—felt creepy approaching after school. System: Tuesdays/Thursdays, 5-8 students in office, paid to do homework; available for feedback on features/messaging—sped up improvement. At Granola, remote Zoom interviews suit meeting product; standing user interviews 4 days/week, anyone joins.
Secret: no need to hear same issue from 10 people—if prototype button confuses one and you see why, fix immediately. Be skeptical; never ask if they'd use (ignore answers). Ask "what next?" then probe negatively: "Is that true? Wouldn't you be tired?" Ignore positives, use conservative metrics: user = new meeting >5 min transcription that day (not prior use or just viewing). Avoid excuses.
Talk to users for context, then use intuition/vision. Granola vision: simple, minimal, nice to spend time in, not distracting. Visualize hundreds of users like "What would Nancy think?" Prioritized user needs lists fail—features unused. Iterative: build minimum, test, fix quickly. Intuitions improve but never stop testing or drift off.
Two paths for AI products: replace human or augment/enhance. Granola augments: starts with meeting notes, expands to post-meeting work (follow-up emails, memos), prep for meetings, analysis across meetings, improving life daily.
Keep EO in your library
Save the videos and channels worth coming back to, and find them again in one place.
