Granola CEO：会议笔记只是开始，真正的机会是重塑工作界面Granola CEO: Meeting Notes Are Just the Start—The Real Prize Is Reinventing the Work Interface
The Takeaway: 我们仍处在计算革命的极早期，会议笔记不是终点，真正的战场是AI原生世界里人们如何协作与思考的界面。
Chris（CJ Pedrigal），Granola联合创始人兼CEO，这家估值15亿美元的AI会议笔记应用曾是最早让人惊呼“这真能用”的产品之一。他坦言创业始终是刀光剑影：无论成功还是挣扎，都像在浪尖上拼命不掉下去。团队从十几人扩到60-70人，最大挑战是如何在规模化中保持产品“灵魂”——一致的体验与品味。他们采用类似Basecamp Shape Up的阶段法：先广泛探索解决方案形状，再小范围验证，最后才做可靠可扩展。竞争方面，Notion、OpenAI、Zoom都跟进了类似功能，但几乎没影响增长，因为“人们今天在争的东西不重要，前面有巨大机会”。
核心策略是成为会议相关场景的世界第一，同时把上下文通过API/MCP开放给Codex、个人Agent等任何地方。他们会预生成数百万份会前简报（即使只有一小部分被打开），因为忙碌用户只有15秒窗口需要它——像楼梯扶手，平时隐形，关键时必须立刻支撑。Chris说：“Meeting notes are not the end all, be all value that everyone's running after. There's something much bigger.” 他认为当前还没有适合处理多上下文、异步代理任务的正确UI隐喻，这是下一个发现而非发明的空白。
对内部，他们刚开始追踪token消耗（因Fable爆发），但产品侧仍优先探索杀手级体验，成本可稍后优化。约一半周活用户以“代理式”方式使用（复杂多步查询），即使他们不这么称呼。
Chris（CJ Pedrigal），Granola联合创始人兼CEO，这家估值15亿美元的AI会议笔记应用曾是最早让人惊呼“这真能用”的产品之一。他坦言创业始终是刀光剑影：无论成功还是挣扎，都像在浪尖上拼命不掉下去。团队从十几人扩到60-70人，最大挑战是如何在规模化中保持产品“灵魂”——一致的体验与品味。他们采用类似Basecamp Shape Up的阶段法：先广泛探索解决方案形状，再小范围验证，最后才做可靠可扩展。竞争方面，Notion、OpenAI、Zoom都跟进了类似功能，但几乎没影响增长，因为“人们今天在争的东西不重要，前面有巨大机会”。
核心策略是成为会议相关场景的世界第一，同时把上下文通过API/MCP开放给Codex、个人Agent等任何地方。他们会预生成数百万份会前简报（即使只有一小部分被打开），因为忙碌用户只有15秒窗口需要它——像楼梯扶手，平时隐形，关键时必须立刻支撑。Chris说：“Meeting notes are not the end all, be all value that everyone's running after. There's something much bigger.” 他认为当前还没有适合处理多上下文、异步代理任务的正确UI隐喻，这是下一个发现而非发明的空白。
对内部，他们刚开始追踪token消耗（因Fable爆发），但产品侧仍优先探索杀手级体验，成本可稍后优化。约一半周活用户以“代理式”方式使用（复杂多步查询），即使他们不这么称呼。
The Takeaway: We are still in the very earliest steps of a computing revolution; meeting notes are not the endgame—the real prize is inventing the interface for how we collaborate and think with AI.
Chris (CJ Pedrigal), cofounder and CEO of Granola, the $1.5B AI meeting-notes app that was one of the first to make people say “holy shit, this actually works.” He describes startups as permanent knife fights: whether things are working or not, you are always on the edge of your abilities, surfing a wave you desperately try not to fall off. Scaling from a 12-person previous startup to 60-70 people has forced hard questions about how to keep product soul—coherent taste and consistent feel—while adding people. They use a Shape-Up-inspired process: shape many possible solutions for a job-to-be-done, validate with a few users, only then harden for reliability and scale. Competition from Notion, OpenAI and Zoom launching similar features barely dented growth; “what people are fighting for today doesn’t matter. There’s this incredible opportunity ahead.”
Strategy: be 5x better than anyone at meeting-adjacent jobs, and make Granola context the easiest fuel for any agent via improving API/MCP. They pre-generate millions of pre-meeting briefs even if only a small percentage open, because the user has a 15-second window of need—like a handrail that is invisible until you trip. “Meeting notes are not the end all, be all value that everyone's running after. There's something much bigger.” He sees a clear gap: no good UI yet exists for working across many contexts and async agent tasks; the right metaphor will be discovered more than invented.
Internally they only started tracking token spend last week after Fable usage spikes; product-side they remain deliberately cost-insensitive while hunting for killer experiences. Roughly half of weekly users already engage in multi-step “agentic” queries even if they would not use that word.
查看原文 →
Chris (CJ Pedrigal), cofounder and CEO of Granola, the $1.5B AI meeting-notes app that was one of the first to make people say “holy shit, this actually works.” He describes startups as permanent knife fights: whether things are working or not, you are always on the edge of your abilities, surfing a wave you desperately try not to fall off. Scaling from a 12-person previous startup to 60-70 people has forced hard questions about how to keep product soul—coherent taste and consistent feel—while adding people. They use a Shape-Up-inspired process: shape many possible solutions for a job-to-be-done, validate with a few users, only then harden for reliability and scale. Competition from Notion, OpenAI and Zoom launching similar features barely dented growth; “what people are fighting for today doesn’t matter. There’s this incredible opportunity ahead.”
Strategy: be 5x better than anyone at meeting-adjacent jobs, and make Granola context the easiest fuel for any agent via improving API/MCP. They pre-generate millions of pre-meeting briefs even if only a small percentage open, because the user has a 15-second window of need—like a handrail that is invisible until you trip. “Meeting notes are not the end all, be all value that everyone's running after. There's something much bigger.” He sees a clear gap: no good UI yet exists for working across many contexts and async agent tasks; the right metaphor will be discovered more than invented.
Internally they only started tracking token spend last week after Fable usage spikes; product-side they remain deliberately cost-insensitive while hunting for killer experiences. Roughly half of weekly users already engage in multi-step “agentic” queries even if they would not use that word.