Notion
A connected workspace for docs, projects and AI agents, built around blocks.
Notion won with a tiny core (blocks, slash commands) even while a rival polished editing, and its community playbook of creator templates carried Notion AI users past the first 24 to 48 hours. Its first assistant shipped before ChatGPT, but it now has about six overlapping automation primitives. Lesson: protect a simple core, seed creators, and give agents connected context.
- Polished markdown-style editor
- Spent years perfecting it
- Added features when it fell short
- Terrible first editor
- Couldn't select between blocks
- Block idea mattered more
Blocks won: a good core beats polish
From Max Schoening interview
What was said about Notion
1Blocks: the tiny core that made Notion
Great products have one tiny core that is exceptionally good. Max Schoening names Notion blocks and slash commands alongside iPhone multitouch, the GitHub pull request, Heroku's git push and Dropbox's menu bar sync icon. Adding one more feature to a weak core never makes a product great.
His advice is to identify the core and push everything else back, as Dropbox should have kept to the single job of files always being there. Quote: "All the great products have something tiny that is a superpower."
Notion's rough block editor beat a competitor's polished one. Max's 2014 startup, a Notion competitor, spent years perfecting a markdown-style editor and added features whenever it fell short. Notion's first editor was terrible, with blocks you could not even select between, but the block idea mattered more and won.
He calls the failure "working diligently on the wrong thing for way too long." Once you find a good core, do not give up on it; iterate until it clicks with users.
Let overlapping features grow, then reconcile them into one core. Notion now has about six automation primitives, including agents, and customers roast it for that. Max says the hard work is reconciling them into the simplest core that should outlive the others, even if the next ship slips.
Fewer features with strong combinatorics beat a high feature count. He cites Claude desktop with three tabs as an industry example. Quote: "Feature count is the same silly metric as lines of code or tokens consumed."
2Designers and PMs who prototype in code
- 1Drop static Figma mockupsCouldn't show how AI feels
- 2Build LLM-friendly playgroundLeast scary, most one-shottable codebase
- 3Prototype all chat thereOnly hurdle: fear of the terminal
- 4Contribute to productionAs models improved, designers and PMs
- 5Ship styling changes directlySmall changes go straight to production
From Max Schoening interview
Notion moved AI prototyping out of Figma because chat mockups are dead. Designing chat interfaces as static Figma images could not show how an AI feels, so two designers, Max included, built a small LLM-friendly playground codebase and moved all chat prototyping into it.
It was built to be the least scary and most one-shottable, using tools LLMs are good at, so people only had to get past fear of the terminal. Notion's main codebase carries a decade of patterns and is not always agent friendly. Max cites Bret Victor's talk Stop Drawing Dead Fish.
The playground was a first step toward designers touching production code. As models improved, the same designers and PMs began contributing to the production codebase, to a lesser degree, and small styling changes can now go straight to production.
Designing in code strains Figma-based processes: marketing still asks designers to rebuild code prototypes in Figma for video assets, which Max calls busywork. At GitHub, about 10% of top contributors were designers even before LLMs.
Code teaches designers the material; shipping is secondary. Max would hire a designer who understands how an agent loop works over one who only tweaks styles in Codex or Claude Code, because agent loops can only be understood by building them in code.
Quote: "To become a master of the material, not a sort of cog in the delivery mechanism for the idea." Production contributions tend to follow once someone is awakened to the medium.
Push non-engineers to terminal tools rather than GUI wrappers. Max encourages PMs and designers to use Claude Code or Codex in the terminal. Designers at Notion mostly use developer tools like the terminal and Conductor, and PMs are slowly adopting them.
The terminal is scary at first, but once people are in it they get curious, pull at other threads and learn more about how computers work.
3Agency, taste and shipping culture on Notion's product team
Hire and develop for agency, not role definitions. Max argues that with skills now at everyone's fingertips, agency, the belief that the world is malleable, decides who gets ahead. Staff who keep asking what a PM or designer is supposed to do will struggle.
Eric Lou asked if Max would hire him in a future startup; told not in the first ten, he moved from strategy docs to Figma to building prototypes to reach the first five. Brian Lovin blurs design and engineering and became a top recruiter because he decided the org needed hires. Max's test: do you drive Notion like it's stolen.
Agency comes from making things, not maneuvering around your boss. The common thread Max sees in high-agency people is making, from a chair to a home-cooked meal, which teaches that the world is built by people no smarter than you.
Taste grows the same way: through rapid reps of ideas and feedback, like training a model, and it cannot be speedrun. Notion names conference rooms after the Macintosh, a Porsche 911 and the first typewriter to surround people with tasteful objects.
Only make obviously good stuff, but iterate in the open. Notion's internal motto is to only make obviously good things, the way the first iPhone and ChatGPT were. The mistake is hiding in a cave until it is good; Max's core value is incremental correctness and more shots on goal.
For mature products he reminds teams not to be precious: "Our users are going to be upset if we don't innovate more so than if we accidentally break a thing." He did this at GitHub and at Notion.
Token spend is not a success metric, though Notion leaves it unlimited. Some individuals at Notion spend thousands, maybe tens of thousands, of dollars on tokens, and spend stays unlimited because exploration is worth it for now.
Max compares bragging about token spend to bragging about lines of code and expects companies to ask about ROI within 6 to 12 months; that is opinion, not data.
Review your own launches as the user would. Max uses jobs to be done to check what users would hire the product for versus what the team wants them to want. A recent Notion feature landing page was rewritten by imagining explaining it to a friend at a whiteboard.
Quote: "Be that user for a second. Would you even buy the thing that you just made?"
4Notion AI and agents
A connected workspace gives Notion's agents the context they need. Notion's first assistant shipped before ChatGPT. Max says agents need context and dislike narrow paths to other data, and Notion's connected workspace lets them roam much like the Unix-like environment coding agents use.
Notion also invested in hard pieces like automatic permission handling in enterprise search. Max says he is surprised at how good Notion's agent turned out.
Creators taught mainstream users Notion AI and cut early churn. For the 2023 Notion AI launch, Notion reused its wiki-era community playbook, relying on YouTube, TikTok, Instagram and Discord creators publishing templates and tips.
Phil Carter says this drove organic adoption and got more new users over the activation hump so they did not churn in the first 24 to 48 hours. Mainstream users are less fluent with AI tools than tech workers assume.
5Lessons others draw from Notion's early days
Buy the .com early; Notion took years to migrate to its. Edward Tian of GPTZero cites Notion's slow move to its .com as a reason to buy the .com early even if unused. GPTZero's .com was available in January 2023 and later was not.
Quote: "if you want to be venture scale you have to buy the .com even if you don't use it"
A modular tool needs a specific value prop for each audience. FLORA's Weber Wong compares explaining his modular creative tool to Notion's early days and learned to describe its value very specifically for each kind of creative team.
Quote: "poetry is about expressing yourself, whereas copywriting is about expressing to someone else"
AI era skills: Why cultivating agency matters more than job titles | Max Schoening (Notion)
Designers and PMs prototype in an LLM-friendly playground codebase and now touch production code. Its first assistant shipped before ChatGPT, and its connected workspace gives agents context; it has about six automation primitives that need reconciling, and unlimited token spend for now.
Buy the .com early; Notion took years to migrate to its. Edward Tian of GPTZero cites Notion's slow move to its .com as a reason to buy the .com early even if unused. GPTZero's .com was available in January 2023 and later was not.
Quote: "if you want to be venture scale you have to buy the .com even if you don't use it"
Notion's lost years, near collapse during COVID, staying small to move fast, building horizontal
Started as a developer tool nobody wanted, then hid its Lego-for-software vision inside a productivity tool. Stayed small with no salesperson until over $10M ARR, and half of B2B customers come from personal users.
Great products have one tiny core that is exceptionally good. Max Schoening names Notion blocks and slash commands alongside iPhone multitouch, the GitHub pull request, Heroku's git push and Dropbox's menu bar sync icon. Adding one more feature to a weak core never makes a product great.
His advice is to identify the core and push everything else back, as Dropbox should have kept to the single job of files always being there. Quote: "All the great products have something tiny that is a superpower."
Notion's rough block editor beat a competitor's polished one. Max's 2014 startup, a Notion competitor, spent years perfecting a markdown-style editor and added features whenever it fell short. Notion's first editor was terrible, with blocks you could not even select between, but the block idea mattered more and won.
He calls the failure "working diligently on the wrong thing for way too long." Once you find a good core, do not give up on it; iterate until it clicks with users.
Let overlapping features grow, then reconcile them into one core. Notion now has about six automation primitives, including agents, and customers roast it for that. Max says the hard work is reconciling them into the simplest core that should outlive the others, even if the next ship slips.
Fewer features with strong combinatorics beat a high feature count. He cites Claude desktop with three tabs as an industry example. Quote: "Feature count is the same silly metric as lines of code or tokens consumed."
Notion moved AI prototyping out of Figma because chat mockups are dead. Designing chat interfaces as static Figma images could not show how an AI feels, so two designers, Max included, built a small LLM-friendly playground codebase and moved all chat prototyping into it.
It was built to be the least scary and most one-shottable, using tools LLMs are good at, so people only had to get past fear of the terminal. Notion's main codebase carries a decade of patterns and is not always agent friendly. Max cites Bret Victor's talk Stop Drawing Dead Fish.
The playground was a first step toward designers touching production code. As models improved, the same designers and PMs began contributing to the production codebase, to a lesser degree, and small styling changes can now go straight to production.
Designing in code strains Figma-based processes: marketing still asks designers to rebuild code prototypes in Figma for video assets, which Max calls busywork. At GitHub, about 10% of top contributors were designers even before LLMs.
Code teaches designers the material; shipping is secondary. Max would hire a designer who understands how an agent loop works over one who only tweaks styles in Codex or Claude Code, because agent loops can only be understood by building them in code.
Quote: "To become a master of the material, not a sort of cog in the delivery mechanism for the idea." Production contributions tend to follow once someone is awakened to the medium.
Push non-engineers to terminal tools rather than GUI wrappers. Max encourages PMs and designers to use Claude Code or Codex in the terminal. Designers at Notion mostly use developer tools like the terminal and Conductor, and PMs are slowly adopting them.
The terminal is scary at first, but once people are in it they get curious, pull at other threads and learn more about how computers work.
Hire and develop for agency, not role definitions. Max argues that with skills now at everyone's fingertips, agency, the belief that the world is malleable, decides who gets ahead. Staff who keep asking what a PM or designer is supposed to do will struggle.
Eric Lou asked if Max would hire him in a future startup; told not in the first ten, he moved from strategy docs to Figma to building prototypes to reach the first five. Brian Lovin blurs design and engineering and became a top recruiter because he decided the org needed hires. Max's test: do you drive Notion like it's stolen.
Agency comes from making things, not maneuvering around your boss. The common thread Max sees in high-agency people is making, from a chair to a home-cooked meal, which teaches that the world is built by people no smarter than you.
Taste grows the same way: through rapid reps of ideas and feedback, like training a model, and it cannot be speedrun. Notion names conference rooms after the Macintosh, a Porsche 911 and the first typewriter to surround people with tasteful objects.
Only make obviously good stuff, but iterate in the open. Notion's internal motto is to only make obviously good things, the way the first iPhone and ChatGPT were. The mistake is hiding in a cave until it is good; Max's core value is incremental correctness and more shots on goal.
For mature products he reminds teams not to be precious: "Our users are going to be upset if we don't innovate more so than if we accidentally break a thing." He did this at GitHub and at Notion.
Token spend is not a success metric, though Notion leaves it unlimited. Some individuals at Notion spend thousands, maybe tens of thousands, of dollars on tokens, and spend stays unlimited because exploration is worth it for now.
Max compares bragging about token spend to bragging about lines of code and expects companies to ask about ROI within 6 to 12 months; that is opinion, not data.
Review your own launches as the user would. Max uses jobs to be done to check what users would hire the product for versus what the team wants them to want. A recent Notion feature landing page was rewritten by imagining explaining it to a friend at a whiteboard.
Quote: "Be that user for a second. Would you even buy the thing that you just made?"
A connected workspace gives Notion's agents the context they need. Notion's first assistant shipped before ChatGPT. Max says agents need context and dislike narrow paths to other data, and Notion's connected workspace lets them roam much like the Unix-like environment coding agents use.
Notion also invested in hard pieces like automatic permission handling in enterprise search. Max says he is surprised at how good Notion's agent turned out.
How Notion’s Community Helped it Win in AI
Won with AI by shipping AI Writer fast and leaning on its community. Grew 10x to over 100 million users, split about a third each across North America, Europe and Asia-Pacific, with at least 25% of business workspaces starting as personal ones. About 80% of customers are consumers but about 80% of revenue comes from businesses. Started as AI Writer, prototyped in a hotel room at an offsite and shipped in an alpha within months. Its word of mouth was very high but repeat use was hard, so Notion built Q&A and database generation into the flow of work.
Creators taught mainstream users Notion AI and cut early churn. For the 2023 Notion AI launch, Notion reused its wiki-era community playbook, relying on YouTube, TikTok, Instagram and Discord creators publishing templates and tips.
Phil Carter says this drove organic adoption and got more new users over the activation hump so they did not churn in the first 24 to 48 hours. Mainstream users are less fluent with AI tools than tech workers assume.
How Notion leveraged community to build a $10B business | Camille Ricketts
Grew through community-led growth: ambassadors starting in 2019 with 20 vocal users, creators, champions and about 30 events a month. It has 400+ people and a valuation near $10B, and a WSJ story and Product Hunt launches were big breaks. Its main metric was monthly net new website visitors.
A modular tool needs a specific value prop for each audience. FLORA's Weber Wong compares explaining his modular creative tool to Notion's early days and learned to describe its value very specifically for each kind of creative team.
Quote: "poetry is about expressing yourself, whereas copywriting is about expressing to someone else"
Edward Tian on Building GPTZero, Scaling to $30M+ ARR, and the Superhuman Acquisition
Took years to migrate to its .com domain, which Edward cites as a reason to buy the .com early even if you don't use it.
The AI Growth Playbook for Subscription Apps – Phil Carter, Elemental Growth
Leaned on YouTube, TikTok, Instagram and Discord creators posting templates and tips for Notion AI in 2023. Community carried users past the activation hump so they did not churn in the first 24 to 48 hours.
This Simple Note-Taking App Makes $20K / Month (and How You Can Build Your Own)
The AI-built demo app sends every note to a Notion board, keeping all notes searchable there while jotting them quickly on an iPhone.
How FLORA is Reinventing Creative Design with AI
Cited as a modular tool that likely learned early it needed a very specific value prop for each audience, a lesson FLORA applied.
An inside look at Deel’s unprecedented growth | Meltem Kuran Berkowitz (Head of Growth)
Meltem cites its late billboard campaigns as well timed, because people already recognized the product and the ads showed its distinctive interface. This is her own hypothesis about why the timing worked.
The ultimate guide to Martech | Austin Hay (Reforge, Ramp, Runway)
A B2B2C example with a user growth funnel at the top that also sells to businesses. It runs HubSpot plus Salesforce, which has no good answer for mapping data.
The ultimate guide to product-led sales | Elena Verna
Notion did not hire its first salesperson until past $10 million in ARR.
Thinking beyond frameworks | Casey Winters (Pinterest, Eventbrite, Airbnb, Tinder, Reddit, Grubhub)
It wrote Casey's bio with a Google Trends start that never happened, and about half was made up. Lenny found his own bio about 70% wrong.
10 lessons on bootstrapping a $200m business | Patrick Campbell (ProfitWell)
Patrick uses it for documentation as one of his tools that cut screen time and batch work.
Lessons from Airtable’s unconventional growth strategy | Zoelle Egner
Its templates feel like users pushing out a beautiful creation, not promoting Notion.
An inside look at Mixpanel’s product journey | Vijay Iyengar
Its pages and blocks are the model for many features hanging off a few strong building blocks. Mixpanel keeps its bets database, with one templated page per bet, and sorts customer feedback in Notion.
How to price your product | Naomi Ionita (Menlo Ventures)
Used as the contrast to Evernote: it leveraged multiplayer growth that Evernote could not, since collaboration lifts acquisition, retention and monetization together.
More in Productivity
ProductivityGrammarly
A writing assistant that checks grammar, tone and clarity across apps and browsers.
10 sources · 43 tips
ProductivityOpal
A screen time app that helps people block distracting apps and reduce phone use.
3 sources · 36 tips
ProductivityDropbox
A cloud storage and file sharing service for individuals and teams.
11 sources · 27 tips
ProductivityGamma
An AI tool for making presentations, documents and websites without fixed slides.
4 sources · 39 tips
ProductivitySuperhuman
A fast, keyboard-driven email app for high performers, now with AI features and team collaboration.
10 sources · 29 tips
ProductivityGmail
A free web-based email service from Google with searchable storage.
7 sources · 11 tips
ProductivityCalendly
A scheduling tool that lets people book meetings by sharing a link to their availability.
5 sources · 22 tips
ProductivityEvernote
A note-taking app positioned as a personal second brain across devices.
4 sources · 10 tips