👋 Howdy to the 4,112 new legends who joined since our last edition! You are now part of a 542,201-strong tribe outperforming the competition together.

LATEST POSTS 📚

If you’re new, not yet a subscriber, or just plain missed it, here are some of our recent editions.

💸 Unfiltered: The Long Game At BILL. How BILL quietly became the back office for 8 million businesses.
🎈 The Unexpected AI Hero (Newsletters). How to convince AI you are the best product on the market.
✨ My Public Speaking Journey. How to wield words like a champion and convince people of your plight.

PARTNERS 💫

Framer helps teams design, build, and launch their marketing sites lightning fast. With the ability to publish hundreds of CMS pages in a single click, operate at a global scale with seamless localization, and even host unified content across multiple domains, teams have never been able to ship faster.

Trusted by companies like Miro, Bilt, and Perplexity.

Join us virtually at Celebrate ‘26 on September 30 at 9:00 am PST to see how leading revenue organizations are using AI to work smarter, stay ahead of the competition, and win more. Don't miss the biggest conversations shaping the future of revenue.

_{Interested in sponsoring these emails? See our partnership options here.}

HOUSEKEEPING 📨

Next week I’m speaking at the annual Startup People Summit about all things culture in startups. You might think I am nothing more than a meme-dealing techno-realist newsletter writer, but in my spare time (50+ hours a week, ha) I’m actually a founder with an organization of ~65 people and engagement scores averaging 93%.

For context—and yes, this is a flex—an engagement rate of 80% across your org is considered excellent. But it’s not been without a bucketload of learnings and screw-ups along the way. So join me next week, while I spill the secrets of what I’ve learned and how we’ve managed to maintain 93% engagement with a full remote team spread across the globe. Sign up here!

ZERO TO ONE 🌱

Granola: Healthy Breakfast & Killer AI Notepad

“Life's this game of inches. One half-step too late or too early, and you don't quite make it. One half-second too slow, too fast, you don't quite catch it. The inches we need are everywhere around us. On this team, we fight for that inch. On this team, we tear ourselves and everyone else around us to pieces for that inch.” This was the famous speech from Coach D'Amato, played by Al Pacino, in Any Given Sunday. This way of thinking typifies startup wunderkind Granola.

Before understanding what makes Granola tick, I had one question rattling around in my head: ‘What is this team doing that makes them so successful?’ I mean, impeccable execution should only get you so far? But the AI notepad, Granola, and its insane rise can really be broken down by a furious approach to being better in every way than the products that came before it.

Not in a Mad Max Fury Road type fury, but more of a meticulous, methodological approach. The scalpel rather than the anvil, if you will, with surgical precision that laid the groundwork for their rise to $1.5 billion within two years of launching.

To unearth what led to their success, I sat down with founders Chris Pedregal and Sam Stephenson, as well as COO Michael Altom. So let’s dive in, folks, to the story of the one and only Granola.

A note-taking journey through time

Before moving ahead, I think we should take a quick look in the rearview mirror. We have clear evidence that humans have been taking notes since time immemorial. The oldest known painting (or possibly note), a warty pig on a cave wall in Sulawesi, Indonesia, is ~45,000 years old. We don't know exactly what the artist was trying to say, but I like to think it was ‘action item: hunt pig.’

If you don’t count that as a true note, I get it, but what about the Ishango Bone, a baboon fibula from 20,000 years back? Humanity's first spreadsheet, carved into a dead monkey's leg. Next stop was Mesopotamia, where Sumerian scribes jotted down grain sales, beer rations, and who owed what on literal wet clay. From there, things moved rapidly.

Fast forward, though, to the 1980’s, and we’d reach what administrative clerks the world over would call the Post-it Revolution. Then the rise of the PalmPilot > Evernote > Apple Notes > and a graveyard of to-do apps that were all branding, no retention.

We’d finally make our way from the caves to the jungles, to the office, and then finally to the remote revolution, and to 2016, when we had the incredible idea of having a robot join our meeting and take the notes for us.

‘Tools for Thought’

Our co-founding duo had a different lead into their startup partnership. Chris had come from a career spanning Google (Gmail, Search, Maps, Google Now), into Socratic (10M users and 2017 Apple App of the Year), a startup he founded that would be acquired by Google, to Google’s in-house incubator to work on a product called Stack, to another acquisition of said product, by you guessed it…Google (Drive).

“I started playing with LLMs about eight months before ChatGPT launched, so a little before people were really talking about it,” he would tell me. “And I became convinced that all the tools we use for knowledge and productivity work were going to change because of LLMs.”

Armed with this unique insight and a voracious fire in his belly, Chris’s next step was to find someone who could come in and help him on the build side. “I needed a co-founder, and the hard parts were either actually training the models or the design side. So it was going to be a researcher or a designer.” Enter: Sam.

Sam and Chris didn’t exactly bump into each other in a bar, nor was their pairing pure happenstance. Chris courted Sam like an old-school romantic love story. He knew what he wanted in a founder relationship, and he went out to find it. What he wanted in this case was someone who could think through this problem from more than just a UI perspective, so he went straight to the ‘Tools for Thought’ community, of which Sam was a part.

You see, a note can be one single, solitary data point filed away in a filing cabinet. This is how most note-takers work today. Sam and his Tools for Thought brethren—and sistren—would think about a note very differently: a note is not just a note, but rather a node in a complex system of notes, sharing context between one another.

The Tools for Thought crowd is obsessed with this stuff. They believe a good notebook doesn't just store your thinking; it makes you smarter, the same way a bicycle doesn't just carry you; it makes you faster. They argue about note-taking the way other people argue about football teams. "It was clear to us that this technology was going to change the world,” said Sam. “But it also felt like no one had figured out an interface for how to make it useful or approachable to real people yet.”

Chris, at long last, had found his startup soul mate: a cold email led to a few cheeky beers, which then led to a Christmas-coded prototype, and the prototype shortly after led to Chris doing the honorable thing, dropping to one knee and asking Sam, “Will you be my co-founder?”

Six to nine months in the wilderness

Once they had struck their romance, it was time to start building, which began with months of open-ended interviews. ‘Walk me through yesterday at work’ surfaced the pattern: meetings were a misery, everyone had tried an AI note-taker, and everyone had churned. The notes were too poor; they felt invasive. We’ve all been there, right?

"Either you spend a lot of cognitive effort trying to take notes, and you're not fully present, or you're fully present, but then tomorrow you're like, 'Crap, what was that thing we talked about?'” as Chris would put it.

Chris and Sam both knew there must be an easier, better way to take notes. And from that, a deeper way to relate back to those notes. But for months, their prototypes bombed.

First came a real-time command UI (’capture this,’ ‘summarize that’), but it failed because it took too much of the user's mental space. Second was somewhat similar: a tab-to-autocomplete version where you would type a scrappy word, hit tab, and Granola would complete it. “It felt kind of like sci-fi to use,” Sam would say. “We were like, 'This is cool.' But, as soon as we put it in people's hands, it became apparent that it wasn't it.” Again, too much mental load.

This led to a lot of soul-searching. The first thing to show signs of life was a web page that transcribed your mic, where you typed one word, and an LLM expanded it into a full note. Each new note deleted the last one. Per Sam, “It wasn't usable as a note-taking app, because if you overwrite the last note every time, that's kind of useless.” But it was the first ‘ooh, there's something new and interesting here’ moment for the team. They knew they were onto something.

A game of inches

Chris and Sam, knowing they were onto something, now had to get it into people's hands while removing all the other pain points in the failed note-taker category. Our preceding ‘game of inches’ idea might inform this section well. As Chris put it, "There wasn't one big decision. There were maybe six or seven smaller product decisions that I think pushed it above some threshold."

Here are a few of the key design decisions Granola took that have, to this point, made it a runaway success.

1/ Desktop app, not hidden tab

One of Chris's beliefs was that the ICP for Granola was the person “running late to back-to-back meetings.” Therefore, the team knew they needed to be a desktop app rather than a tab you can never find. “When you're three minutes late with 50 tabs open, the app is findable, and the tab isn't.”

2/ Kill the bottttttttttts

This is key. Everyone has joined a meeting where humans are outnumbered by bots in some sort of final-stage Butlerian Jihad dystopia. It’s weird; the bots take up half the screen; everyone feels like their privacy has been wholly invaded. "A weird third wheel on a date," as Chris so aptly put it.

Granola works on the system audio instead of needing to join the meeting, so it works in Zoom, Meet, Huddles, and even in person. No more needing to integrate, and no more creepy faceless weirdo annotating your dates.

3/ No video, no stored audio

Another important feature for the paranoid amongst us: Granola stores only notes and a transcript. It’s your smart notepad, not a surveillance Big Brother state app, salivating for a Palantir partnership. (They tried to kill the transcript, too, actually, but users found it too useful.)

4/ Guest-specific notes built for you

Granola also researches who you are, and "does a little internet research on you and figures out who you are,” per Chris. For the same meeting, the VC and the founder walk away with completely different notes: ‘Not a fund returner,’ and ‘I think that went really well’ type stuff, but a heck of a lot smarter.

5/ The meeting notification

Right as you are scampering around your desktop, trying to get organized for your next call, a lovely little notification pops up in the right corner of your screen. I call this Granola’s ‘crisis averted’ button. One click and, like magic, it opens Zoom (or Meet/Riverside/the ghost of Skype) and Granola together. No new habit required; it slots into the panic you already have.

6/ The year-long wait to launch

And lastly, the guys moved really intentionally early on. They onboarded two users a day, watched how they used the tool, fixed bugs, and repeated the process. They leaned into polish before publicity.

In addition to the six points above, there have been thousands of other micro-improvements to Granola as a tool. Now… I know what you're thinking? You’re thinking, ‘Where is the life-changing innovation?’ Well, let me direct you to my old friend Al ‘Coach D’Amato’ Pacino, and his 4:31 sermon on the topic.

See you in five…

Do you get it now? It’s not that there aren’t any cancer-curing features of this AI notepad; the entire point is that there aren’t any. Life's this game of inches. And on this team (Granola), they will fight for that inch. And when we add up all those inches, that's going to make the fucking difference between winning and losing!

_{Legal fine print: The ’inch by inch’ speech remains the intellectual property of Warner Bros., Oliver Stone, and the ghost of Coach Tony D'Amato. Quoting Al Pacino at 4:31 in the morning is commentary, not plagiarism, and my lawyer (I don't have a lawyer) assures me this is fine.}

The (best) worst customers in the world

Once Granola had a working product, they needed to get it in the hands of users. And their first bad idea—a point countless people would reiterate to them—was actually one of their greatest assets in disguise. The contrarian ICP bet they made was backing investors to love the product. Everyone said it was ICP suicide. ‘Don't build for VCs; they are too few, too weird, too cheap, too slow.’

VCs, operationally speaking, are a wonderful fit for a product like Granola. Back-to-back meetings, switching context between a seed-stage climate tech pitch and a Series B fintech, and so forth. VCs require the cerebral flexibility of an eight-year-old Russian gymnast just to survive the day.

The sneaky upside when betting on VCs, though, is that if the product works, you have an in-built distribution machine. Venture capitalists are inherently the best connected people in tech. They spend their days rubbing Patagonia vests (I know that’s more a Wall Street thing, but humor me) with the most exciting startups, scale-ups, and enterprises known to man.

"Everyone told us it was a terrible idea,” Chris said. “Turns out it might have been bad advice, because it was the best thing we ever did.” This virtuous cycle of goodness meant there was no need for a marketing budget until their Series B (more on that soon): "It was all word of mouth, man. Word of mouth and Twitter, basically... We did nothing growth-hacky. It's almost anti-growth-hacky.”

🔥 Hidden hack: Build a product investors genuinely love, and raising money becomes a hell of a lot easier.

Selling a product that sells itself

As a startup founder, there are usually a million problems on your plate at all times. ‘Is anyone going to use this product? How do we make payroll this month? Can we community adjust this EBITDA?’ are a few of the typical questions you’ll ask yourself building a startup. ‘How on Earth are we going to service all this inbound?’ is less common, and what you’d probably call a first-world problem. A problem most founders would sell their left airsack (lung) for.

This is the problem incoming COO Michael would face upon joining in 2025. Granola had never spent a dime on paid marketing and had a sales team of one, yet they were growing like a rocket. “I was astonished by the quality of inbound leads coming in so early in the journey with nothing in place,” says Mike. “We didn't have SOC 2 at the time. We didn't have a lot of the enterprise features. There were basically a lot of reasons not to buy Granola at the enterprise or company level 18 months ago.” Knowing that the company was in somewhat of a free-hit position (incredible product-led growth), Mike set out to build a GTM system that would add fuel to the Granola fire. These were the initial steps Mike would roll out.

The first step Michael needed to take was to build out the marketing function. The question was, “Can we accelerate the growth of the product even more by creating a lot more awareness outside the niche Granola was famous in?” Given PMF was there, the idea was to test the awareness side, investing in podcasts and newsletters (like this one).

_{*Note: This piece and surrounding Granola campaign actually came to be because Open Source CEO is Michael’s favorite newsletter. From what I understand, he basically forced us onto his marketing department. A beautiful gesture.}

These would help spread the word far and wide, and hopefully get bottom-of-the-funnel users to start getting their hands on Granola, leading to further product-led growth opportunities. The next stage would be converting this top-of-funnel and usage into enterprise accounts.

To convert from the bottom up, Michael’s next step was to move from one sales all-rounder to a fully fledged local sales team. I say local because the first Granola sales team was based with boots firmly on the ground—likely Doc Martins, considering the Granola office's proximity to famous Brick Lane—in London.

Granola’s sales team was five at the time, but would be answering demand from its largest market, the U.S., long into the night, so it didn’t take long for Michael and the team to decide to open a San Francisco-based office to properly serve North America. Something, along with hiring their first VP of Sales, Bardia, that would happen only three short months later.

”By September of last year, I had enough signal,” Michael told me. “We chatted with the founders first, then with the board, and we agreed to build a much bigger and more structured sales team, not just to answer inbound but to be much more proactive. That’s when we made the decision to open an SF office.”

Switching verbatim for a moment…

Doc: “What does it actually take to sell on top of PLG?”

Michael: “In a typical sales-led process, you don't have any users in your target buyer already using the product. In most cases, you identify your ICP, you identify your target accounts, and you give those accounts to your team. It's very theoretical: ‘I think my ICP is healthcare companies on the West Coast.’ The strategy is to find the best accounts that match your ICP, prioritize them, give them to SDRs and salespeople, and tell them to go for it.

In a PLG motion like Granola's, especially when the product is so bottoms-up and so viral within companies, using product data is really helpful, because you're not just shooting in the dark. It's not that one day I woke up and decided healthcare is the vertical to go after. Here, I follow the PLG. We can really look at which companies and which domains have users using the product. Do they retain well in that domain? Do they share invites and notes internally? Do we see some sort of virality or adoption? Are they using more advanced product features? We can use all of those signals to prioritize, so we're not being theoretical. We follow where we feel we have the best chance to capture the demand.”

Still Michael: “That means we can really help reps know which accounts to target and when, because we have a lot of signals. It's like, ‘Okay, looks like something's happening in this company; they've been sharing like crazy in the last week; we really need to reach out.’ And we can identify exactly who inside that company are the power users, the people sharing within the company. Those folks are much more open to responding to our outreach, because we're trying to help them, in a way.”

Which brings us to the marketing budget, now that one finally exists. After some experimentation, Michael’s team spends roughly half of their resources on top-of-funnel awareness spend—podcasts, newsletters, out-of-home—that is deliberately, almost proudly, hard to measure.

With a product that is as close to viral as can be, the game for Granola is finding top-of-funnel channels that they like, and backing them until they can prove or disprove their hunch. "You need to keep the conviction,” Michael would tell me, when looking back on their quest to crack podcast advertising as a growth channel, “And look for the data points that make you feel better about it.”

The Five Golden Rules of AI Products

Chris isn't really a manifesto guy, but in preparation for this deep dive, I heard him lay out his five rules for building AI products, and it's the closest thing Granola has to a written-down doctrine. I'm reprinting them here because Granola is the case study for every single row. Here they are.

Rule #4 is the one worth sitting with, because it's the real answer to the question hanging over this whole piece: what's stopping Google from just...doing this? If you ask Chris, maths, mostly. AI costs scale per user. Granola can afford to run frontier models over every set of notes for a few million users; Google Drive would have to do the same for a few billion. The copycats cloned the interface fast enough. They can't clone the unit economics of being small.

Five rules, though, don't make people tattoo your logo on their calendar. For that, Granola leans on something older than AI…

Never underestimate ‘a little magic’

Sometimes, in order to get someone's attention in this life, you need to hit them with one of the most underrated elements in all of technology, product, and UI. Something that’ll get them to straighten their spine and sit upright: magic.

When I think of creating a little magic moment, I can’t help but think of the great Michael 'Squints' Palledorous (b. 1950); a ”smart aleck kid who wears glasses with thick black frames and dons a plain black hat,” says Google. To me, Squints was more than some smart aleck kid; he was a role model.

If you don’t know Squints, please take a short 3:28 seconds out of your day to watch this clip.

I will admit, the Squints scene from The Sandlot has very little overlap with our story today. But when I think of ‘magic moments,’ that scene just pops into my head. Another magic moment, and one I wrote about recently, is Attio's CRM onboarding flow.

As Sam would put it: “The animation feels like clever shit is happening right here, right now.” And it really feels like that. If you remember your first ever experience with Google Earth, it kinda feels like that. ‘Ohh, this is what the future looks like,’ type vibes.

Other companies I think could boast this effect are: 1) the first time you sent a Loom, 2) same goes with your first Calendly link, and 3) when you installed Grammarly for the first time and became a true word wizard.

This might sound trivial; it’s not. Viral, word-of-mouth growth only happens when you blow people away. Take NPS scores for example; it’s hard to get a single figure from the Internet, but it's directionally true that if you score a 7-8 in an NPS, you will tell zero people about your experience; score a nine, and you’ll tell five people; land a perfect score of 10, and you will tell 9-10 people. This, ladies and gentlemen, is the ticket to viral growth, and a touch of magic plays a huge part.

Granola’s final boss: Context

Something I noticed throughout the research for this piece was that Chris, Sam, and Michael all had a pretty clear vision of what Granola is today, and what Granola can be tomorrow. The vision, which is pretty evocative and is being attacked from a number of players in the AI space, is the full context layer for an organization.

There is a world in which you can see Granola being the first-party owner of said context, and it doesn’t take one too many steps to get you there: Granola joins from the bottom up, spreads through the organization, better context begets more usage begets better context, and so on.

I was discussing this with a friend who posited the question: “Why won’t Google just do that?” Which is a fair question. Google, though, only joins Google calls; Granola joins Google Meet, Teams, Slack Huddles, in-person. Hell, I think it’d join Skype if it ever came back from the dead. Granola, in the future, can and should tie into every tool you use on a daily basis, and one that adds a rich layer of contextuality to all of them.

Playbook / how you can apply this

• Win on inches, not haymakers: Granola had no killer feature. Six or seven small decisions compounded past the ‘I'll use this daily’ threshold. Audit your product for missing inches, not missing moonshots.

• Enter hated markets, not empty ones. Note-takers had existed for seven years, and everyone who tried them churned. Universal demand plus universal disappointment beats a blank space every time.

• Subtract to differentiate: No bot, no video, no auto-emails, no growth hacks. In a copycat market, what you refuse to build can be as powerful as what you do.

• Pick the ICP with the loudest mouth: VCs were a ‘terrible’ customer and a perfect distribution channel. Choose early users for where they carry you, not just what they pay you.

• Kill your sci-fi darling: If the demo wows but the usage doesn't, it's theatre. Granola lost a year to live AI notes before letting go.

• Ride the rising tide: If your product doesn't get better every time the models do, you're cooked. Skip the problems the next release will solve for free.

Future

What you notice with Granola when you zoom out far enough is that this was a love story all along. Not only was it the co-founding love story of Chris and Sam, but also the love story that the users of Granola have with the product, and even the love the team has for this particular problem.

Here's the thing about notes that Granola figured out before the rest of us: they were never the point. "Notes aren't the point of a meeting," Chris told me. "The point of a meeting is usually that you make some decisions and then have to go and act on them." That's where this is headed: the follow-up email drafted before you ask (now live), the CRM updated, your prep already waiting when you join the call. This is what we all really want deep down. Beauty, love, Granola.

Fun fact

• Granola, a healthful food born out of a trademark war: Granola the food started as ‘Granula,’ invented by health reformer James Caleb Jackson in 1863. John Harvey Kellogg nicked the recipe, got threatened with a lawsuit, changed exactly one letter, and trademarked ‘granola.’ A century and a half later, one letter's worth of legal creativity is worth $1.5 billion.

Extra reading / learning

• AI note-taking app Granola raises $43M at $250M valuation, TechCrunch - May, 2025

• Investing in what makes us human, Index Ventures - March, 2026

• Granola raises $125M, hits $1.5B valuation, TechCrunch - March, 2026

And that’s it! You can follow Chris, Sam and Michael on LinkedIn, and also check out Granola on their website to learn more.

BRAIN FOOD 🧠 

TWEETS OF THE WEEK 🐣 

TOOLS WE RECOMMEND 🛠️

Every week, we highlight tools we like and those we actually use inside our business and give them an honest review. Today, we are highlighting Lightfield*—an AI-native CRM that assembles itself from your email, calendar, and meetings.

See the full set of tools we use inside Athyna & Open Source CEO here.

HOW I CAN HELP 🥳

P.S. Want to work together?

1. Hiring global talent: If you’re hiring tech, business or ops talent and want to do it 80% less, check out my startup, Athyna. 🌏

2. See my tech stack: Find our suite of tools & resources for both this newsletter and Athyna here. 🧰 

3. Reach an audience of tech leaders: Advertise with us if you want to get in front of founders, investors and leaders in tech. 👀