Feb 15, 2024 · 42m · no-priors
No Priors Ep. 51 | With Notion CEO Ivan Zhao
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
Notion co-founder and CEO Ivan Zhao discusses with hosts Sarah Guo and Elad Gil how generative AI, semantic retrieval, and modular building blocks are dismantling SaaS fragmentation and transforming modern knowledge work.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 28.1% of the talking time here. How this is scored →
The hosts as informed peer 5.1 Guest teaching 5.2 Guest disagreement 1.1 The hosts pushing back 0.5
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 8:57 Refuting the premise of user app-building
Ivan directly contradicts Sarah's question by stating people do not wake up wanting to build applications, explaining why Notion had to pivot toward ready-made templates.
Hardest push from the hosts ▶ 27:26 Elad counters Ivan's macro bundling explanationElad pushes back on Ivan's historical cycle thesis, offering an alternative argument that bundling is driven by shifts in founder product quality philosophy and cross-selling strategy.
Biggest teaching moment ▶ 24:53 Romance of the Three Kingdoms and computing wavesIvan synthesizes classical Chinese literature with forty years of PC and SaaS software history to educate the hosts on why AI models inherently necessitate data bundling.
The host holds their own ▶ 29:05 Sarah analyzes Oracle's database bundling playbookSarah demonstrates deep enterprise software knowledge by breaking down Oracle's historical M&A and database lock-in strategy to contextualize modern LLM embedding spaces.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Breaking SaaS Fragmentation and Computing as Universal Literacy | 4 | 5 | 1 | 0 | Sarah opens asking why more SaaS companies do not pursue unified data models. Ivan provides a philosophical overview tracing computing back to universal literacy and early pioneers, framing software creators as modern factory workers. | |
| The Transformative Potential of Retrieval-Augmented Generation | 6 | 5 | 2 | 1 | Ivan playfully pushes back on Sarah's multi-decade timeline, noting RAG is already transforming computing today. Elad demonstrates technical familiarity by characterizing the new paradigm as probabilistic software engineering. | |
| Douglas Engelbart's Influence and the Evolution to Templates | 5 | 6 | 3 | 0 | Ivan directly counters Sarah's premise by asserting that users do not actually want to build software from scratch, explaining Notion's pivotal realization that building blocks had to be pre-packaged into practical templates. | |
| The Breakthrough of GPT-4 and Betting the Company | 4 | 5 | 0 | 0 | Sarah inquires about resource prioritization around AI. Ivan candidly recounts sleeping on GPT-3 before experiencing an inflection point with GPT-4's reasoning capabilities, prompting a bet-the-company pivot. | |
| Technical Trade-offs, Human Inertia, and Incremental Innovation | 5 | 6 | 1 | 1 | Elad asks what foundational capabilities are currently missing. Ivan educates on design philosophy, citing Virgil Abloh and Steve Jobs on why software must balance novel capabilities with incremental changes to overcome human inertia. | |
| Instant Knowledge Retrieval with Notion AI Q&A | 4 | 6 | 0 | 0 | Sarah asks Ivan to explain Notion AI Q&A. Ivan details how RAG replaces lexical search and eliminates friction from routine workplace queries and asynchronous message ping-pong. | |
| AI Agents, Smart Calendaring, and the Future of Communication | 6 | 5 | 1 | 1 | Elad bridges RAG with agentic workflows in calendaring. Sarah contributes an insightful synthesis about Notion AI fundamentally reducing the necessity of human-to-human coordination messages. | |
| Semantic Retrieval and Moving Beyond Manual File Organization | 5 | 7 | 2 | 0 | When Sarah asks whether users still need structured organization, Ivan firmly argues semantic retrieval makes manual filing obsolete. He frames current AI dynamics through computing history and classical literature. | |
| Distribution Bundling Versus Information Bundling in Enterprise Software | 8 | 4 | 2 | 4 | Elad offers an alternative thesis on bundling driven by founder psychology and product philosophy, citing Rippling and HubSpot. Sarah draws a historical parallel to Oracle's database suite, prompting Ivan to distinguish distribution bundling from data bundling. | |
| Notion's Design Philosophy and Cultivating Designers Who Code | 6 | 5 | 1 | 0 | Sarah asks about Notion's design culture. Ivan explains why centralized decision-making and cross-disciplinary designers who code enable unique trade-offs, while Elad reminisces on Ivan's early pitch and aesthetic clarity. | |
| Office Culture, Design Classics, and Physical Environments | 6 | 4 | 1 | 0 | The hosts discuss office culture, furniture classics, and multi-model architectures. Ivan suggests that future UI paradigms might even discard Notion's iconic left navigation sidebar in favor of pure retrieval. | |
| The Separation of Front Office Workspaces and Back Office Verticals | 4 | 6 | 0 | 0 | Sarah asks about AI opportunities outside Notion's scope. Ivan maps the market by differentiating horizontal front-office workspaces from deeply verticalized back-office systems like Snowflake or specialized legal tools. | |
| Enterprise Cost Consolidation and Expanding Market Scale | 3 | 4 | 0 | 0 | Sarah wraps up by asking about enterprise adoption in a tightened macroeconomic climate. Ivan emphasizes that consolidation provides immediate cost savings for CFOs alongside operational efficiency. |
Statements from this episode (19)
Zhao: No-code originated from developer customization, not data unification
“No code is like coming from this kind of like power user developer angle of wouldn't be nice. Everybody can modify this underlying software they use every day. That's why I go there. It wasn't coming from the angle of the knowledge and data wants to be in one …”
Zhao: SF is expensive because it is the world's software factory
“Because there's a separation between people who can make software and people who use software. That's why SF's rent is so expensive, because we're the modern day Detroit or Manchester, right? We're the factory of the world.”
Zhao: Notion historically had almost no machine learning engineers
“We have somewhat a small team of search, but we don't have too many ML folks, almost nothing.”
Zhao: Prompt engineering effectively makes anyone a real-time machine learning engineer
“Language prompt engineer language model sort of make everybody like a real-time machine learning engineer.”
Ivan Zhao: Modular software tools must be packaged as templates
“Our learning and pivot is instead of giving people those Software building tool, we have to package the software building blocks together as ready to use templates, as ready to use use cases that people can adopt really quickly.”
Zhao initially overlooked GPT-3 because it seemed limited to marketing copy
“I personally, I have to admit, I slept on it. On GPT-III, even saw GPT-III, it feels like, what is this thing useful for? It's like, yes, for marketing, for content writing, for creative first draft. Didn't really click for me.”
Zhao: SaaS is digital paper-pushing that LLMs can now perform
“Fundamentally, SaaS, software is all, we're all in the same information, people, paper pushing activity, right? It's like a piece of paper coming in front of you. A human, like, changed a couple bits, pushed to another human. Language model can do some form of…”
Zhao: Notion bet the company on AI after seeing GPT-4
“And after that, we sort of just bet the company on it.”
Zhao: AI applications must stay largely familiar to overcome user inertia
“You cannot make something too new. You have to be largely the same and change one thing and two things, right? Virtual ablo. The off-white guys, like, three percent difference. Just push the boundaries so people can accept it, but still also new, right? To me,…”
Gil: RAG and Embeddings Are Under-Appreciated Relative to Impact
“I feel like RAG and embeddings are very under-discussed, or under-appreciated in some sense, relative to the impact that they really seem to be having, or starting to have, and I think Notion Q&A is a great example of that.”
Zhao: People Will Communicate Less as AI Agents Handle Exchanges
“I probably feel it's probably less the agent side essentially bet on language model, that's the communication.”
Zhao: Embeddings and RAG are eliminating the need for manual organization
“I think organization might be, we might be moving away from the organizational world. Why do you need to organize because you can retrieve? Why do you have index? Like index initially are file cabinets and little index are sitting on top of so you can find thi…”
Zhao: AI Language Models Are Driving the Next Software Bundling Wave
“Now the new technology happening is AI language model, and if you build more with it or just think more with it, language model wants information to be one place. Once the endpoints to be connected, so it's easier to, is it hard enough to like a conversion lan…”
Gil: Bundlers Like Rippling Win Without Covering Every Feature Edge Case
“Now I feel like people are both building great products as well as bundling them, but also they're much more aggressive about saying it can be 80% as good. It could be 50% as good, but I'm going to have a bundle and that's HubSpot and that's Rippling and they …”
Guo: Oracle's Strategy Was Buying Second-Best Products and Leveraging Distribution
“If you look at the strategy of companies like Oracle, right? It was very much for a decade and a half, like a dominant, at least commercially attitude of like, okay, we're going to buy the second best product in this additional software category we want to be …”
Zhao: LLMs Demand Bundling of Information Rather Than Just Distribution
“I think there's bundling of distribution and bundling of information. What you're describing to me is more of Microsoft, more like bundling of distribution. Language model wants the bundling of data, bundling information.”
Zhao: 80% of Notion designers know how to code
“In notion with most of our designer can code. Majority of, 80% of the designer can code”
Zhao: Hybrid designers and engineers make superior product trade-offs
“The moment you can, as a designer or as an engineer, you can code, you can design, you can make really interesting trade-offs, right? At the end of the technologies, at least in my opinion, is about trade-offs. What kind of trade-off you can make that unlock n…”