Ivan Zhao | Long Strange Trip w Brian Halligan Summary

1. Key Themes

The "Jazz Band" Company: A New Organizational Operating Mode Beyond Founder Mode

Notion has consciously rejected the "marching band" model of hierarchical execution in favor of a "jazz band" model — where talented individuals improvise together with shared sensibility rather than following rigid top-down instructions. This is now being identified as a distinct evolutionary step beyond "founder mode."

"We started a mantra in the company. We want to be a jazz band, not a marching band... I realized I cannot be a marching band person. That's just not me. I will feel like if everybody, I delegate everything, everybody do things. What do I do? I feel like I can't do that. So it's like a self-reflecting value thing. It's auto equilibrium with me. So it's like, okay, I'm a jazz band person. So we've been hiring more and more jazz band person. Also AI happened in the past two and a half years. So more jazz band people can really shine during this moment." 00:20:22 — Ivan Zhao

The AI Refounding Imperative: Feel It, Don't Read About It

Ivan's core argument is that any company — startup or mature SaaS — must actively build with AI to understand what paths open up, not just study it from the outside. The model introduced a fundamentally new "technology-first" development paradigm replacing "customer-first."

"Building with language model back then, and somewhat still is, it's like brewing beer. You can't truly predict the things... You sort of have to throw your best people in there, see what the model, see what technology gives you. So it's not customer first. It's more like experiment with this technology first." 00:06:29 — Ivan Zhao

"If you feel, you realize most tech industry is support different flavor knowledge work. We're at the beginning of how this can do due to this industry... You have to feel it. You have to build it. Building is one of the best way for your product or build for your internal systems. Build for yourself. Yeah, tinker on a weekend, tinker on a side." 00:37:50 — Ivan Zhao

The Talent Equation Has Been Rewritten: Capability Is Commoditized, Taste and Agency Are Not

Notion's hiring philosophy has fundamentally shifted away from experience and credentials toward taste (value system + aesthetic judgment) and agency (will + drive). This is a direct consequence of AI democratizing raw capability.

"A talent equal to this person's capability, such experience, times this person's taste or value system, times this person's agency will. What language model changes, like just like Google, everybody can access information, language model allow everybody to be a pretty good writer and programmer. So capability got normalized, democratized. And taste becomes still important... And will, how hard do you work? This you cannot change. So we're optimized for the latter two today." 00:11:40 — Ivan Zhao

The Barbell Engineering Org: Super Junior + Super Senior, Cut the Middle

Notion is deliberately building a barbell-shaped engineering organization — hiring fresh, high-agency junior engineers alongside highly experienced senior architects — and eliminating the expensive middle layer of experienced-but-not-exceptional senior engineers.

"Hire for super junior, super senior... Because the senior provide taste. There's still a lot of auto distribution information out in language model, architecting and language models, no one very bad with that. So they need to provide a direction. It's almost like this, a good engineer managing four to six coding agents at once, right? But a good, really senior architectural engineer can managing two or three junior interns or engineer, they each manage two or three, then that's the number is higher, right." 00:12:53 — Ivan Zhao

The Circle Replaces the Triangle: AI as Organizational Steel

The historical org chart is a triangle that distorts information. The new model is circular — with AI context at the center — and represents a structural upgrade as significant as steel was to architecture.

"The analogy I like to use, what you've been using somewhat is like, before steel, buildings cannot be higher than five, six floors... After steel, there's information where waste structure can go much higher. Since sometimes language model plus software is that. It's the steel for organizations." 00:09:35 — Ivan Zhao

2. Contrarian Perspectives

Contrarian: Don't First-Principle Your Sales Motion — Respect the Playbook

Most technically-minded founders believe they can redesign go-to-market from first principles. Ivan tried and explicitly calls it a mistake. The modern enterprise sales playbook exists because of deep human psychology, not legacy inertia.

"We tried to put our innovation point onto sales back then. Just go classic sales. It works... You should only have, each company should only preserve your innovation point to a few places, go a couple places. You cannot spread everywhere. Like, otherwise, you're spread too thin. You're trying to reinvent the wheel too many times." 00:54:53 — Ivan Zhao

"A lot of things you want to talk to a seller. There's a lot of the reason the modern sales playbook has been around for two-ish decades... You want to see a human, you feel comfortable. You don't want to be a doctor from a robot. No. Then why should you buy an expensive thing from a robot? It's not that hard, I would say." 00:54:12 — Ivan Zhao

Contrarian: Gross Margins Will Get Worse — and That's the Right Move

Conventional SaaS wisdom protects gross margins obsessively. Ivan argues that willingness to let gross margins compress while running experiments with AI infrastructure is a requirement of being serious about the transition — not a warning sign.

"Oh, you have to. You're, you're down for that. Like, what are you doing? Then you're not in the war mode." 00:23:33 — Ivan Zhao

Contrarian: Acqui-hiring Founders Is a Decalcification Strategy, Not Just a Talent Strategy

Most companies treat acqui-hires as a talent play. Ivan frames them explicitly as an anti-calcification mechanism — a way to continuously inject disruptive energy into a company that would otherwise rigidify as it scales.

"We intentionally do a lot of acquisitions. We have 50, 60 founders at Notion... Founders are kind of just kind of like little decalcified meathead machinery just trying to break things, right? So keep your regenerating." 00:38:59 — Ivan Zhao

Contrarian: Non-Founder CEOs May Not Be Able to Refound

In an era where every SaaS company is being told it must transform into an AI company, Ivan quietly argues that this transformation may be structurally unavailable to non-founder-led companies.

"I don't know. It would be tough. I think founder has the moral high ground to change things. People are more tolerant and forgiving with founders, right?" 00:37:21 — Ivan Zhao

Contrarian: It's Getting Harder to Start a Company, Not Easier

Against the prevailing narrative that AI lowers the barriers to startups, Ivan argues that while starting is easier, meaningful scaling is harder than ever because incumbents are shipping faster and every niche attracts immediate competition.

"It's never been easier to start. Man, it's never been harder to scale because like in every little micro segment, if there's a little traction, like there's a hundred competitors two days later." 00:41:06 — Brian Halligan

"You can argue that it might be more difficult to start new companies today compared to even a year ago. Because the world's getting more noisy. Like the shipping cadence from existing companies companies are just increasing." 00:40:08 — Ivan Zhao

3. Companies Identified

Notion Description: All-in-one knowledge management and productivity platform Why mentioned: Core subject of interview; canonical example of a SaaS company successfully refounding as an AI company, maintaining small team discipline, building with language models from early on, and pioneering new organizational structures

"In my mind, he's the canonical example of how a SaaS company can move and become an AI company." 00:00:56 — Brian Halligan

GitHub Description: Developer platform owned by Microsoft Why mentioned: Notion's CRO Erica came from GitHub as CRO, bringing enterprise sales system-thinking credibility to Notion's go-to-market transformation

"Our CRO, Erica, she was CRO of GitHub before. It's a system thinker, right?" 00:56:08 — Ivan Zhao

Anthropic Description: AI safety company and model developer Why mentioned: Built a custom model for Notion during the early AI product phase — notable that even with Anthropic's direct involvement, the early AI products still didn't work, illustrating how early the technology was

"Simon, I would say my problem and Simon's problem, we might be living too much into the future. It bites you with new technology. Like we try everything. Entropic built a model for us. We work open AI fine tuning or another model, just none of them work." 00:35:33 — Ivan Zhao

Rippling Description: HR and workforce management platform Why mentioned: Parker Conrad cited as a parallel example of a founder who builds cross-functional rather than siloed — held up as a model for how to prevent organizational information loss

"I interviewed Parker Conrad from Rippling and he's similar as a lot of founders... The way it kind of strikes me is like the bigger HubSpot got almost all the important things cut across all those organizations." 00:19:57 — Brian Halligan

4. People Identified

Ivan Zhao Description: Co-founder and CEO of Notion Why mentioned: Central subject; described as the "canonical refounder" — twice rebuilt his company from near-zero, first by moving to Kyoto with his co-founder, second by pivoting to AI in Cancun after early GPT-4 access. Pioneering a new organizational philosophy built around jazz-band culture, taste-first hiring, and AI-native workflows.

"The best artists reinvent themselves. Miles Davis has three, at least three faces, right? Picasso two to three faces... Steve Jobs, the king of refounding." 00:25:04 — Ivan Zhao

Erica (Notion CRO, former CRO of GitHub) Description: Chief Revenue Officer at Notion Why mentioned: Identified as the key hire that finally made Notion's enterprise sales motion work — specifically for being a "systems thinker" who could complement a high-energy sales leader

"Our CRO, Erica, she was CRO of GitHub before. It's a system thinker, right? We pair Erica with our head of sale, Pravesh, which is rah, rah, like the meat eater... This is when things start to work in the past year and a half." 00:56:08 — Ivan Zhao

Pravesh (Notion Head of Sales) Description: Head of Sales at Notion Why mentioned: Paired with Erica as the complementary "rah-rah meat eater" to her systems-thinking CRO role — held up as an example of the right pairing to make enterprise sales work

"We pair Erica with our head of sale, Pravesh, which is rah, rah, like the meat eater. Banes in his teeth. Perfect pairing." 00:56:37 — Ivan Zhao

Douglas Engelbart and Alan Kay Description: Pioneering computer scientists from the 1960s-70s, inventors of foundational human-computer interaction concepts Why mentioned: Ivan laments that most tech founders don't know these figures, arguing that this historical ignorance weakens the industry's capacity for genuine innovation

"If we ask people who Douglas Engelbar or Alan Kay, those computing pioneers are, most people in tech have no idea." 00:31:27 — Ivan Zhao

Ray Ozzie Description: Creator of Lotus Notes, former Microsoft CTO Why mentioned: Halligan's former boss who was obsessed with Engelbart and Kay — represents the broken lineage between early computing humanists and today's tech culture; also positioned as proof that knowledge management, once a "shit industry," is now central and valuable

"I worked for a while with this guy, Ray Ozzie... We were in the knowledge management industry, which is a shit industry because chief knowledge officers have no money and it was never urgent. In theory, it sounded great. Now knowledge management is actually incredibly important and works incredibly well." 00:44:40 — Brian Halligan

5. Operating Insights

The CMO Org Should Be Split in Two

Notion eliminated their CMO organization entirely and restructured marketing into two distinct units: one sitting next to product focused on storytelling and social, the other dedicated purely to sales/demand gen. The insight is that a CMO sitting between both functions creates a costly information round-trip that slows both down.

"We no longer have a CMO organization... We break up our CMO organization into storytelling, which is closer to product... because the product is changing so fast... And then the other part is like serving the sales go-to-market function, like demand gen, lead gen. Rather than information round trip to a CMO, then figure out how to serve both the product side and the gold markets, no more." 00:15:47 — Ivan Zhao

First Interview for Sales Reps: Build Something, Don't Review Resumes

Notion changed the first step of their sales hiring process — instead of resume review, they ask candidates to build something and send a Notion link. This screens for AI fluency and agency before any other evaluation.

"We changed our first interview. No longer look at your resume, build something for us, send the notion link, send something that you build. We look at what you..." 00:17:36 — Ivan Zhao

Separate Financial Planning From Product Planning Completely

Notion operates with a strict separation: financial planning is marched conservatively (you need to know your running pace), while product strategy has no plan at all by design, because the market and technology change week by week.

"Financial planning is still, at least I find useful to give you a... you know, you're on track. Maybe we know how fast you're running... But product strategy is all, there's no freaking plan. No plan. Because what changes the market, the technology is so fast. It's not in the six months. It's not three months. It's week by week. So this is, you have to jazz this thing." 00:22:05 — Ivan Zhao

Use a Teleprompter + Speech-to-Text Workflow for All-Hands Prep

Ivan's practical technique for all-hands as an introvert with English as a second language: speak your thoughts the night before into a speech-to-text model, then use the output on a teleprompter. This decouples thinking from speaking under pressure.

"The night before, or then I just like talk to myself and got what I'm roughly want to say. And the time doing all hands, teleprompters there." 00:47:24 — Ivan Zhao

6. Overlooked Insights

Knowledge Work Is Only 150 Years Old — We Are Free to Reinvent It Entirely

This was mentioned almost in passing near the end of the conversation, but it is a profound reframe. Most people treat the structures of modern work as if they were immutable features of human civilization. Ivan's point is that knowledge work is a recent, constructed invention — which means it is fully available for reinvention, and AI may represent the first genuine redesign in a century and a half.

"We always forget that modern knowledge work is only about 150 years old. It's invented. It's not as old as fire or language. It's only 150 years old. So why cannot be a new flavor of it?" 00:45:19 — Ivan Zhao

This isn't just philosophical. For investors, this is the bull case for why Notion — and platforms like it — aren't competing in an existing category. They are potentially defining what work looks like for the next 150 years. The total addressable transformation is far larger than any current SaaS market sizing would suggest.

Chinese Open-Weight Models Are Already Good Enough for Enterprise Knowledge Work

Ivan briefly and matter-of-factly mentioned that for the bulk of Notion's use cases — which involve routine knowledge work rather than frontier coding — they can already substitute cheaper Chinese open-weight models instead of frontier models. This has dramatic gross margin implications and is a signal that for the knowledge work segment specifically, the commodity model layer is arriving faster than most assume.

"For knowledge work, you, you can, we already start to see this. You sort of can settle with the, the second year, the Chinese models, the open weight ones. They're not super smart. Like a lot of this kind of paper pushing information in your company. I spill coffee on the carpet. Can HR team come someone to tidy this up next week? That doesn't require Opus to file tickets." 00:23:58 — Ivan Zhao

This is a non-obvious competitive signal: companies building AI products for knowledge work (vs. coding/reasoning) may achieve meaningfully better unit economics than the market currently prices in — and the window to capture this before competitors notice may be short.