How Granola Grew to $1.5B: The Growth Strategy Behind the AI Notepad
How Granola used a year-long beta, VCs, founders, shareable meeting notes, and bottom-up adoption to grow the AI notepad into a $1.5B company.
Granola grew by starting with a small but unusually well-connected group of venture capitalists, spending about a year refining the product with roughly 100 to 150 beta users, and building a meeting-notes product people naturally exposed to coworkers.
The distribution path was simple: investors used Granola on founder calls, founders discovered it, founders brought it into their leadership teams, and polished meeting notes spread the product internally. Granola then added collaboration and enterprise controls after pockets of individual usage were already forming inside companies.
The company launched publicly in May 2024. By October, it reported roughly 5,000 weekly users, 50% retention after ten weeks among an early cohort, and six meetings per week among retained users. In May 2025, CEO Chris Pedregal told TechCrunch that the user base had been growing about 10% per week. In March 2026, Granola raised $125 million at a $1.5 billion valuation.
Granola has not disclosed a reliable current user count or absolute ARR figure. Its growth story is better understood through retention, distribution, and the way individual usage turned into company adoption.
Granola growth timeline
| Date | Milestone |
|---|---|
| March 2023 | Chris Pedregal and Sam Stephenson start Granola |
| 2023–May 2024 | The team spends about a year in beta, onboarding roughly 100–150 users and repeatedly changing the product |
| May 22, 2024 | Granola launches publicly as an "AI-powered notepad for meetings" |
| October 2024 | Granola reports ~5,000 weekly users, 6x growth in weekly meetings, and 50% 10-week retention among an early cohort; raises a $20M Series A |
| May 2025 | Pedregal says the user base has been growing roughly 10% per week; Granola launches team collaboration and raises $43M at a $250M valuation |
| 2025 | Granola expands from individual notes into shared folders, team knowledge, cross-meeting chat, Windows, Recipes, and other workflows |
| March 2026 | Granola raises a $125M Series C at a $1.5B valuation |
| 2026 | Granola pushes further into enterprise context with MCP, APIs, governance controls, mobile apps, and company-wide workflows |
Granola's Series A investors were already users
Granola's Series A investors did not need a pitch deck to understand the product. Several had already encountered or used it.
That was partly luck. It was also a consequence of an unusually specific first customer.
Granola initially focused on venture capitalists because they spend much of the day in meetings, often take notes in repeated formats, were relatively easy for the founders to reach, and sit in the middle of a dense network of founders, operators, co-investors, and portfolio companies.
The market was small. The network was not.
An investor could use Granola on a founder pitch, share the notes internally, mention the product to the founder, and expose it to several other people without Granola needing to manufacture a referral loop.
By October 2024, the wedge had already widened. TechCrunch reported that more than half of Granola's users were in leadership positions, while investors had become a smaller portion of the user base.
Granola had used VCs as the entry point without becoming a VC-only product.
The first versions made people watch the AI
Chris Pedregal and Sam Stephenson founded Granola in 2023 after experimenting with several GPT-powered work tools.
The meeting-notes idea did not immediately look like the final product.
One early version generated notes in real time. It was visually impressive, but it created a bad meeting experience: users watched the AI write, judged whether it was correct, and paid less attention to the person speaking.
Another version lived in the browser and disappeared among open tabs. The team responded to weak experiments by adding features and structure, which made the product more complicated without solving the core problem.
Granola eventually moved most of the AI work to the end of the meeting and cut roughly half of the product.
The interaction became much quieter.
A user writes down whatever seems important during the call. Granola transcribes in the background. When the meeting ends, the product combines the user's notes with the transcript and fleshes those notes out.
That difference matters because the human notes become a signal about what the user cared about. The AI is not deciding from scratch which parts of a conversation deserve attention.
Pedregal and Stephenson have discussed the evolution of the product in interviews with Every, The Cognitive Revolution, and other product interviews.
The lesson was not simply "use AI for meeting notes." It was that the most impressive AI behavior in a demo was not necessarily the behavior users wanted during an actual meeting.
Granola spent a year watching a small number of users
Granola did not launch a large waitlist and try to understand product-market fit from aggregate dashboards.
The team onboarded users slowly and watched individual behavior closely.
Founding engineer Jim Fisher later described how Granola used a "dot-plot" instead of relying mainly on conventional dashboards. Each row represented a user and showed the days on which that person had recorded meetings.
A retention percentage might tell the team that usage went down.
A dot-plot showed who stopped using the product.
That produced better questions:
- Why did this person use Granola every weekday?
- Why did another person only use it on Tuesdays?
- Why did somebody suddenly stop after a month?
- Did one user invite colleagues?
- Was a team beginning to form around an existing individual user?
Granola says the dot-plot was effectively the dashboard on its wall for the year before launch. The founders combined those behavioral observations with repeated user conversations.
By October 2024, Granola reported that half of people in an early cohort were still using the product after ten weeks, and retained users were recording about six meetings per week.
Those are company-reported figures rather than audited numbers, but they help explain why Granola could rely so heavily on recommendation and repeated use after launch.
The year-long beta was not a waiting-room tactic. It was the product-development process.
Why VCs were such an effective first market
Pedregal has described the initial-user criteria as people who had many meetings, relatively repeated meeting formats, and were easy for the founders to reach.
VCs matched all three.
But they had another property that mattered even more: distribution density.
A venture capitalist might speak to several founders in a day. They also talk to other investors, operators, and people inside portfolio companies. If Granola became part of that person's meeting workflow, the product could repeatedly appear around exactly the kind of users a new B2B productivity tool wanted to reach.
This is different from picking the largest possible initial market.
Granola selected a relatively small market where:
- the problem happened frequently;
- the user could give high-quality feedback; and
- each user interacted with many potential future users.
That created a natural bridge from VCs to founders.
Pedregal has said that after public launch the team deliberately broadened its focus toward founders rather than continuing to optimize Granola around investors. Founders were a harder and more varied user group, but they also provided a path into leadership teams and companies.
The early wedge therefore did two jobs at the same time: it helped Granola improve the product and placed the product inside a network that could carry it outward.
"AI notepad" was a better category than "meeting bot"
The AI meeting category was already crowded when Granola launched.
Many competitors joined calls as visible bots, recorded the conversation, and generated summaries afterward.
Granola positioned itself differently:
an AI notepad for meetings.
The product runs on the user's device and captures system audio rather than sending a bot into the call. It can therefore work across meeting platforms and in other conversation settings without adding another participant named "AI Notetaker."
Granola's founders have said they never seriously wanted the bot experience. In a 2026 explanation, they argued that bots behave differently across platforms, do not work well for in-person conversations, and can make a meeting feel more awkward.
That choice came with a real growth trade-off.
A visible meeting bot is also free advertising. Everyone in the meeting sees the product name.
Granola gave up that exposure.
Instead, it bet that a better experience would lead users to recommend the product voluntarily.
That decision also made the positioning easier to understand. Granola looked and behaved more like a notepad than enterprise call-recording software: the user still writes; the AI helps afterward.
Product quality made the sharing loop possible
Granola's growth loop only works if the notes are good enough to share.
That sounds obvious, but it is an important part of the story.
Pedregal has discussed a strategy of using strong frontier models even when they were more expensive. A smaller startup can sometimes afford to give a relatively small user base expensive inference that would be much harder to deploy across millions of existing users.
Granola 2.0 later made this philosophy explicit by adding access to top models from providers including OpenAI, Anthropic, and Google, with Granola selecting stronger reasoning models for more demanding analysis.
The exact model stack has changed over time, and claims about competitors' model costs are difficult to verify from the outside. The broader product bet is clearer: Granola prioritized the quality of the artifact a user received at the end of a meeting.
That quality mattered because the output itself became distribution.
If the notes were mediocre, users would keep them private or rewrite them.
If the notes were surprisingly useful, they would send them to everyone who needed the meeting context.
The public launch was smaller than the later story
Granola launched publicly for macOS on May 22, 2024.
The product was presented as a simple AI-powered notepad that listened to a meeting and used the user's rough notes plus the transcript to generate a better version afterward.
The launch included founder posts, a product GIF, Product Hunt, an official announcement, and press coverage.
It did not look like the launch of a future $1.5 billion company.
That is useful context because Granola's growth did not depend on one enormous launch spike.
The product compounded after launch.
By October 2024, TechCrunch reported that Granola had grown its user base about fivefold to roughly 5,000 weekly users. Weekly meetings had increased sixfold. Granola said 70% of people who used the product for a meeting during their first week came back, and its Series A announcement said half of people in an early cohort were still using it ten weeks later.
The more important event was not launch day.
It was thousands of users repeatedly opening Granola every time another meeting started.
The notes became the referral
Sam Stephenson has said Granola's early growth came largely from recommendations and people sharing notes.
The second behavior is the more interesting growth loop.
Imagine a meeting ends.
A participant sends a polished set of notes into Slack a short time later. The notes are far more complete than anybody could have manually written in that time.
Someone asks:
What did you use to make this?
The artifact demonstrates the product before the recipient has ever visited the website.
That is a strong form of product-led distribution because the audience is also unusually qualified. The people receiving a meeting summary are coworkers, collaborators, founders, investors, customers, or other people connected to the meeting.
Granola supported the behavior with shareable links and Slack sending. Granola 2.0 expanded it further with shared folders, links that could be viewed by teammates, cross-meeting chat, company browsing, and automatic Slack posting.
This was not the same as having no growth loop.
Granola avoided forced invitations and a visible bot, but useful notes still carried the product from one person to another.
The difference is that the sharing action had value even if Granola received no new user from it.
People already needed to send meeting notes.
Granola inserted itself into an existing behavior.
Granola's growth loop
The early loop can be simplified to:
VC uses Granola
→ shares or discusses notes with founders
→ founder starts using Granola
→ founder shares notes with leadership team
→ coworkers discover the product
→ several employees begin using it individually
→ a pocket of usage forms inside the company
→ IT or leadership formalizes the account
→ shared company context makes Granola more useful
→ more meetings and notes create more opportunities for discovery
That is why Granola's initial customer choice mattered so much.
The company did not need every VC in the world. It needed enough highly connected users for the product to move into founders and companies.
Collaboration turned personal use into company adoption
Granola's enterprise path looks similar to classic bottom-up SaaS.
One person starts using the product.
Coworkers receive notes and begin using it themselves.
A cluster forms inside the company.
Eventually, IT, security, or management sees enough usage to justify a formal deployment.
TechCrunch reported that companies including Cursor and Gusto adopted Granola after employees had already begun using it individually.
Granola's own Brex case study says the product reached more than 50% of knowledge workers within six months of enterprise adoption. Those numbers come from Granola and should be treated as company-reported, but they illustrate the expansion pattern.
Granola 2.0 was designed around this shift.
Shared folders let teams collect meetings around sales, customer research, recruiting, or projects. Teammates could browse shared context and ask questions across multiple conversations. Slack auto-posting put meeting summaries into a place coworkers already spent time.
By March 2026, Granola had added enterprise features including access controls, SSO, SCIM, APIs, MCP support, consent tools, retention controls, and other governance features.
The enterprise layer did not need to create the first demand.
It gave companies a way to manage demand that employees had already created.
Pricing evolved with the move from individual tool to team product
Granola's pricing changed as its product moved from personal notepad to shared company context.
At launch, Granola offered an individual product with a limited free allowance followed by a low monthly subscription.
By 2025, it had introduced a Business plan alongside the individual product as team collaboration became a larger part of the strategy.
Today, Granola's official pricing is organized around three tiers:
| Plan | Current price | Main role |
|---|---|---|
| Basic | Free | Individual usage with limited history |
| Business | $14/user/month | Full history, integrations, advanced models, and team workflows |
| Enterprise | $35+/user/month | SSO, administration, governance, and API access |
The important strategic change is not a single price point.
Granola increasingly made the valuable parts of the product — shared context, integrations, long-term history, stronger models, and enterprise controls — useful at the team level rather than treating collaboration as an afterthought.
That supported the same expansion path its organic growth had already created.
Granola eventually started marketing like a larger company
It is misleading to describe Granola's entire growth story as "no marketing."
The early period was unusually organic.
Later, the company began behaving more like a scaled consumerized B2B brand: campaigns, paid ads, out-of-home placements, podcast sponsorships, customer stories, referral programs, affiliates, and launch marketing all became part of the mix.
The important sequencing is that paid distribution came after the company had evidence of repeated use and natural sharing.
One example was Granola's 2025 "Crunched" year-in-review experience, which generated personalized summaries from users' meeting histories and was designed to produce outputs people might want to share.
The campaign reused the same basic mechanic as Granola's original growth loop:
make an artifact valuable enough that users voluntarily distribute it.
Granola's later marketing became louder, but the underlying distribution idea remained recognizable.
Granola expanded only after owning the meeting-notes wedge
A pure meeting-summary product faces an obvious long-term problem.
Transcription and summarization are becoming cheaper, and Zoom, Teams, Google, Notion, AI assistants, and dozens of startups can all generate meeting notes.
Granola's response has been to move upward from notes into context.
The product evolution roughly looks like:
meeting notes
→ searchable meeting history
→ chat across conversations
→ shared team folders
→ workflows and Recipes
→ MCP and integrations
→ company-wide context and APIs
Granola 2.0 described conversations as some of the most current information inside a company. Its Series C announcement made the same argument more explicitly: the opportunity is not only capturing notes, but capturing context and making that context useful in other workflows.
That shift matters for growth because it increases expansion potential.
One useful personal notepad can become a team knowledge layer.
A team knowledge layer can become infrastructure that other software and AI agents use.
Granola growth metrics: what we actually know
Granola is now a highly valued private company, but several numbers repeated online are estimates rather than disclosed facts.
Here are the figures with the strongest public sourcing:
| Metric | Figure | Date | Source |
|---|---|---|---|
| Weekly active users | ~5,000 | October 2024 | TechCrunch, citing Granola |
| User growth since launch | ~5x | October 2024 | TechCrunch, citing Granola |
| Weekly meetings growth | ~6x | October 2024 | Granola / TechCrunch |
| First-week return rate | ~70% | October 2024 | TechCrunch, citing Granola |
| 10-week retention | ~50% | October 2024 | Granola |
| Meetings per retained user | ~6/week | October 2024 | Granola |
| User-base growth | ~10% per week | May 2025 | Chris Pedregal via TechCrunch |
| Series A | $20M | October 2024 | Granola |
| Series B | $43M at $250M valuation | May 2025 | Granola / TechCrunch |
| Series C | $125M at $1.5B valuation | March 2026 | Granola |
| Current ARR | Not publicly disclosed | — | — |
| Current active users | Not reliably disclosed | — | — |
Third-party estimates of Granola's ARR or total users may be directionally interesting, but they should not be presented as company-disclosed figures.
That distinction matters if the goal is to understand what actually drove the business rather than reverse-engineer a narrative from an estimated revenue chart.
What Granola did differently
Granola's growth can be reduced to six decisions.
1. It optimized the product before optimizing acquisition
The company spent roughly a year working with a small beta group and was willing to delete major parts of the product before public launch.
2. It chose an initial market based on network position, not market size
VCs were not the ultimate market.
They were high-frequency users sitting directly next to founders, operators, and other potential customers.
3. It rejected the category's obvious viral mechanic
Meeting bots create free impressions because every attendee sees them.
Granola gave that up in favor of a quieter local notepad experience.
4. It made the output itself distributable
Meeting notes already needed to travel between people.
Granola made those notes good enough and easy enough to share that the artifact became acquisition.
5. It used retention to turn word of mouth into compounding growth
A recommendation only matters if the person who tries the product keeps using it.
Granola's early retention numbers suggest that a meaningful portion did.
6. It waited for bottom-up usage before building the enterprise layer
Team and enterprise features amplified existing behavior instead of trying to manufacture enterprise demand from zero.
A small, connected niche can beat a large audience
The most useful lesson from Granola is not simply "word of mouth works."
Most products do not get meaningful word of mouth.
Granola combined several conditions that made it possible:
- a high-frequency problem;
- a product people used several times per week;
- strong retention;
- output that naturally needed to be shared;
- early users connected to the next market;
- and a product experience distinctive enough to talk about.
The initial VC market was valuable because of both usage frequency and network adjacency.
That framework is useful for other startups choosing an early niche.
Do not only ask:
How many customers exist in this market?
Also ask:
How often do they experience the problem?
Who sees the work they produce?
Who do they talk to every day?
If they love the product, where does it travel next?
Granola's first market was small. Its edges touched a much larger one.
Frequently asked questions
How did Granola grow?
Granola grew through a year-long hands-on beta, strong retention, and word of mouth. It initially targeted venture capitalists, expanded toward founders, and spread inside companies when users shared polished meeting notes with coworkers. Collaboration and enterprise features later turned pockets of individual adoption into team accounts.
When did Granola launch?
Granola launched publicly on May 22, 2024, as an AI-powered notepad for meetings.
How many users does Granola have?
Granola has not published a reliable current active-user count. TechCrunch reported roughly 5,000 weekly users in October 2024, about five months after launch.
How fast did Granola grow?
In October 2024, Granola said weekly users had grown roughly sixfold since launch and TechCrunch reported about 5,000 weekly users. In May 2025, CEO Chris Pedregal told TechCrunch that the user base had been growing around 10% per week.
What is Granola's ARR?
Granola has not publicly disclosed an absolute current ARR figure. Third-party websites publish estimates, but those should not be treated as company-reported revenue.
What is Granola worth?
Granola raised a $125 million Series C in March 2026 at a $1.5 billion valuation.
Why did Granola target VCs first?
VCs have many meetings, tend to use repeated note formats, were accessible to the founding team, and interact constantly with founders and other investors. That made them both useful beta users and unusually effective distribution nodes.
Why doesn't Granola use a meeting bot?
Granola captures audio through the user's device rather than sending a bot into a call. The founders say this makes the product work more consistently across online and in-person conversations and preserves the feel of a private notepad rather than a recorder.
Sources
Primary and company sources
- Introducing Granola: the AI-powered notepad for meetings — Granola, May 2024
- Granola raises $20M to build the AI notepad that makes you smarter — Granola, October 2024
- Start with a dot-plot — Granola, May 2025
- Granola 2.0: A second brain for your team — Granola, May 2025
- Granola raises $125M to put your company's context to work — Granola, March 2026
- Why Granola doesn't use a bot — Granola, June 2026
- Granola pricing for startups — Granola, 2026
- Brex saves 427 hours a week with Granola — Granola case study
Interviews and reporting
- Granola debuts an AI notepad for meetings — TechCrunch, May 2024
- VCs love using the AI meeting notepad Granola, so they gave it $20M — TechCrunch, October 2024
- AI notetaking app Granola raises $43M at $250M valuation — TechCrunch, May 2025
- Granola raises $125M, hits $1.5B valuation — TechCrunch, March 2026
- Chris Pedregal on choosing VCs as Granola's initial market — MAD Podcast
- Calm AI for Crazy Days: Inside Granola's Design Philosophy — The Cognitive Revolution
- Sam Stephenson on Granola's product development — product interview
- Granola's Growth Playbook — Startup Riders, April 2026; useful secondary synthesis of founder interviews and historical product/pricing decisions