AI takes over meetings, the SF vs London tech scene, and ethics in media

Support for today's show comes from Anthropic, the team behind Claude. You know how sometimes a problem just grabs you? Like you sit down thinking it's a quick thing and suddenly it's midnight? That's exactly the kind of mind Claude is built for. People who don't just want the answer, they want to chase the thing that's underneath it. Anthropic positions Claude as a thinking partner, not a search engine. It works through the problem with you and it doesn't try to wrap things up neatly, which, if you've spent any time on this show, you know that's a feature, not a bug. Get started with Claude for free at Claude AI Access and see why problem solvers choose Claude as their thinking partner.

0:00

Ugh.

0:36

You said you were over him, but his hoodie's still in your rotation. It's time. Grab your phone, snap a few pics and sell it on depop. Listed in minutes with no selling fees. And just like that, a guy 500 miles away just paid full price for your closure. And right on cue.

0:36

Hey.

0:53

Still got my hoodie?

0:54

Nope. But I've got tonight's dinner paid for. Start selling on Depop, where taste recognizes taste list. Now with no selling fees, payment processing fees and boosting fees still apply. See website for details.

0:55

Ellis has AI Butthole icons in our show prep notes. And I just want Ellis to ask about that. What is that about?

1:10

Obviously, it is different than the little Claude butthole I have or the OpenAI techno butthole. Yeah, I mean, it's certainly adjacent, right?

1:17

Yeah, it's a butthole adjacent. Yeah, I'll give you that. Yeah.

1:24

This week on Access, Alex and I chat about his upcoming field trip to OpenAI, the rising relevance of film, photography and ethics in the modern age of content creation. Then we're joined by Sam Stevenson, co founder of Granola. The viral AI note taking app is hot on the heels of their big raise, which values Granola at more than $1 billion. We get into it all from their controversial rebrand, if AI should actually run our meetings, opening their first office in San Francisco, and much more. Welcome to.

1:28

Welcome to Access. I am Alex Heath here as always with my co host, Ellis Hamburger. Ellis, how's it going?

2:02

Oh, I'm all right. As always, I am afflicted by something from my children, one of whom was up at 4am last night for unknown reasons.

2:08

We'll get into that. I'm pretty excited for this week's guest. I am a granola power user. Did I convert you to granola or Were you already a user as well?

2:19

It's honestly hard to keep track of how many things you've pilled me on. Claude being the latest. I think you need to start getting like a kickback.

2:28

I do. I do need a kickback. I've definitely pilled a lot of people on granola. We have the co founder, Sam Stevenson. It's a great chat. Really interested in the direction they're going, AI note taking, how I'm outsourcing my brain, all of the above.

2:36

Well, but Alex, what is your ethics situation now that you're no longer in a big media newsroom? Are we accepting hats and water bottles now? At the very least.

2:53

Oh, like what? I, you know, accept some granola. Granola? Is that what you're saying?

3:01

Exactly.

3:06

Yeah. You know, if it's perishable, I'll accept it.

3:07

I mean, it is funny. I mean, I love the spirit of a lot of these, you know, rules and ethics statements because it is such a slippery slope. But when you look at it in the moment and it's like if you were to be biased by a sack of granola to like corrupt your entire editorial brand with some corrupted take, that would be pretty, pretty sad.

3:13

My line is something that could reasonably influence my coverage or comes with like a quid pro quo of like, you know, we give you this, you do this. There's obviously situations where a company wants you to come to their event and they're down to pay for travel and there's no strings attached of like how I cover it. And in that case, you know, maybe I would do it. It is kind of company dependent, just since you asked. But I do think I'm thinking a lot about it. I think even with the pod, I think it's important. I think people don't want to be surprised. I think people want to know what is sponsored, what is not. Like you worked with granola. We talk about that in the show. But like, you know, there's obviously that component of like you did some consulting with them, but they're not paying us to do this pod when we have a guest on. I guess, I guess it's a good, good time to talk about this. This wasn't on the to do list. But yeah, when we have a guest on, you know, they're not being paid. They're not paying us to, to come on and be a guest. Right. Even though, like, maybe you've worked with them or in the case of like Ivan from notion.

3:33

Right.

4:37

I've notion sponsored the show when we launched and like that was awesome. We really appreciated that. But that had no bearing on, you know, him coming on the show or what have you.

4:37

I was always mystified back when I was at the Verge. People are like, hey, so do we have to pay you to cover us or something? And I'm like, what. What on earth are you talking about?

4:47

I mean, people expect it. I just think being up front, where

4:58

does that even come from?

5:01

I think everyone assumes everything is pay to play these days. And in the world of influencers and YouTubers and all of that, that that's just the way it is. And, like, we're kind of straddling this interesting line where, you know, you're not a journalist. I still am, but I'm independent. And I do want people to know that, like, the, the conversations we're having on the show are editorially, us, they're controlled by us. They're not like, bought. If something's an ad and we're reading an ad, obviously it's an ad, but, you know, the things are separated and what the. The line I have is like, they're not being a kind of secret quid pro quo of like, you know, if you, if you do the ad, then you have the CEO on, you know, we're just not going to do that.

5:03

We don't make it easy, though, do we? When the show called Access.

5:51

No, but that's a little on the nose, I think, on purpose, when we were talking about it. Right. Like, it's kind of playing with that idea of, you know, it was looked down on. And I mean, it still is, I think, by traditional media, like, quote, unquote, access journalism. But, you know, I think what we're trying to do is just have good, good, funny, interesting, provoking conversations with cool people.

5:54

Well, here's the question. I think some people would say that if you do have the access to these people, it is your, you know, I don't know, moral responsibility to, like, give them shit or hold them accountable. Where do you feel like the state of that is with media right now? Because I feel like with antagonistic media, obviously we're trying to be a different type of show, get to know the founders, hear what they think. Bit more entertainment, speak to my background, your background, our own personal opinions as well. But I think a lot of executives are like, I'm just not even gonna field an email from someone who's kind of antagonistic.

6:16

I think it's bifurcating. I think it's media that is not independent is getting increasingly antagonistic, and then independent media is probably getting a little too soft. I hope we can find some kind of middle ground where, you know, even if it comes up in the form of me saying, like, I think the product doesn't work or whatever, I think. I think what you'll hear if you actually listen and pay attention to our conversations is like, yes, we're. We're not here to, like, grill the people coming on, but we're also asking them, like, good questions. We're asking them questions that, like, their VC would not ask them, and we're just doing it in a much more, I think, approachable way, in a way that kind of disarms them, makes them feel comfortable. I think that's the art of it, is asking good questions not being too differential, but also, like, being a human being and not making it a unpleasant experience for the guest.

6:55

I guess this art is literally your job, is it not? I mean, like, you go home and you scoop their leaked memos, and then you have them on the show and you talk about their buttons in the app. Don't look right.

7:46

I mean, ideally. Ideally, I think there's room for both, and I try to do both, but this is, like, less of the scoopy part of me that you're getting on the show, though. I do love scoops in any context. But, yeah, man, I don't know. I think we're. I think we're figuring it out. I think. I just think if you're coming on a show like this and you're spending an hour of your time to be on a pod, like, to expect that person to be subjected to, you know, intense grilling is just not reasonable. I. I don't know. It's just the. The medium is the message. And I think pods are conversational and hanging out and having fun and learning, but also having fun and in a newsletter or written text format. It's just different. You can get that across more in a way that you can't through pause. I don't know.

8:01

Yeah.

8:45

Well, for anybody listening out there, we would love your thoughts. Feel free to tweet at us.

8:45

Or are we softies at Alex or at Hamburger?

8:50

Yes, at Hamburger. I am not hard to find.

8:54

I'm at Alexey Heath. But first, Ellis, have you recovered? It sounds like. Not based on what you said earlier from having a dozen children partying at your house. What? My wife and I came to your all's lovely birthday party for your daughter over the weekend. You had the bouncy house, you had the cake, you had the film camera, you had the adult Beverages for the adults and the kids. Beverages for the kids. How was it?

8:57

Yeah, it was great. And I think the hashtag tech angle is, you know, just think about how you spend your time at the party. Obviously nobody wants to be the dad who's like permanently behind the Sony Handycam, recording the entire thing. But it's interesting. You go through these phases of your life and when you look back at the memories, they are marked by the app or the hardware you were using at the time to document your life. And I mean, I honestly, as kind of charming as it is, I honestly have major regrets about like documenting my firstborn's first steps on Snapchat at like 360p pixelated next to the dancing hot dog.

9:22

You didn't get that on film. That was just snap.

10:00

I wasn't into the film cameras yet at the time. And then meanwhile, now that I'm shooting more film as a result, that means I'm capturing less video, you know what I'm saying? And it's a trade off, right? And it kind of, as a tech nerd, can be a lot of pressure to have these different phones, cameras, tools around you and just having to decide which one you're going to use to capture the moment and. But yeah, I'm very happily in my film era right now, as many who know me know. And I think what's most interesting about it is that, you know, it's not just me. I think we're seeing a lot of ascendant products, whether it is this new no Fusion app that's going viral, whether it's Halide, the camera app. One of the founders just went to Apple some time ago to help work on their camera app or even my friend Andy Allen's not Boring camera. And the trend right now is to try and make the photos look a lot less digital and a lot less over sharpened, which is literally one of the claims on the no Fusion app store and make them look a bit softer. And I think it's interesting debate in tech. We're so often all about the megapixels and the speeds and feeds. But I think as it turns out, when people look back at their camera roll, they care less about how HDR the moment is and more like, does it actually feel emotional? Does it actually feel like the memory? And you know, this. This certainly coincides with all the Gen Z trends around, getting back toward analog stuff. Some say the trend is overreported, but I feel like everybody I know, when they look at the, the film photos I post, they say how do I do that?

10:03

Oh, yeah, your photos are great. I mean, this is the. This is the AI pushback. Right. Or I guess the reaction to the rise of AI and slop is, like, people crave stuff that looks and feels real, and I think these kind of photos are that. And I don't know, I think it's cool. I think it's cool that you're documenting, you know, your daughter's birthday this way.

11:43

Yeah.

12:05

And there really is kind of ineffable. Ineffable quality to it. It's just kind of sad. I feel like sometimes I look back at iPhone pictures, and I much prefer the ones often that were filtered on Instagram or on Hipstamatic or on Huji Cam or on David Dobrik's Dispo camera, because they also feel like a reflection of where I was at and the tools I was using and the aesthetic I was interested in at that point in my life, you know? Whereas when you look back at the iPhone photos, no matter where you are, what year it is, the sky is always that exact shade of blue that we all know so well. I'll just never forget the, like, computational photography stuff. And the Verge was, like, all over this when Samsung was, like, creating a simulated moon. No matter what picture of the moon you took, they would put the moon on it.

12:05

Yeah. Where I used to work the Verge, that was a whole huge storyline for us. Yeah. What is a photo? And it still is a thing that they talk about a lot. Yeah. I mean, I feel like we're kind of all past that now as a society, though. I feel like I just assume nothing is real. And I think that's why your film photos, you know, resonate for people, is when you see it, it's like your mind just instinctively wants to believe that that's real because of how it looks.

12:53

Yeah. And the apps are getting better at simulating it. Like, Daz in particular, which I found from former Verge alum Becca Farsace, has a particularly good, like, disposable version especially. I think you realize that, like, I don't think it's just a. Like, a simulation of feeling when you use the flash indoors on a camera. Like, I do think it generally does look cool a lot of the time, but no one uses the iPhone flash. The iPhone itself doesn't even want to use the flash ever. And so indoor photos just always have, like, a very specific look, and there's no feeling of, like, being out at the party and being spontaneous. So, Yeah, I don't know. I guess we'll see if it's, if it's a cyclical trend just like everything else. But it's, it's an interesting thing to watch People, people wrangling with that, I would say.

13:17

Yeah. Well, I'm going to download this. No fusion app. This looks cool.

14:06

Yeah.

14:10

Well, what's up with you?

14:11

I'm back on the road, man.

14:12

You're going back to San Francisco this week?

14:13

Yeah. OpenAI invited me out to spend some time with the Codex leadership team and the Claw father, Peter Steinberger, who is now two weeks into OpenAI. So I'm going to be interviewing him and Thibaut, the Codex lead, and some of the research people and we're going to have a dinner at the OpenAI Library. Sushi in the library.

14:15

So how much did they spend on OpenCloud?

14:38

Again? It's not known.

14:40

Aren't you sources news?

14:41

It's not known. I mean they. So the Open Claw is an open source project that is now being maintained by an outside foundation. Dave Morin, OG Facebook platform dude built Path. He is now partner Offline Ventures. He is now running that foundation with, on behalf of Peter externally and it's a nonprofit. Peter joined OpenAI to Essentially, I mean, I'm going to find out, you know, tomorrow when I'm up there, but I think essentially to basically build a chatgpt openclaw, you know, something that is as simple and normie friendly as chat, but as powerful as, as openclaw.

14:43

Well, the reason I bring it up, I mean, it's just like Facebook spent hundreds of billions of dollars on the metaverse rename their company Meta and then that trend just kind of didn't happen. And there's just so much FOMO driven acquisition or aqui hire behavior in the Valley. And it's just been interesting to see whether it is with Facebook's own Manus app that they acquired or Claude adding dispatch recently so that you could do Open Claw y behaviors even when you're not at your computer. It's like it hasn't even been three months since they kind of got that guy. And the whole idea that that's the future. And as it turns out, a lot of this stuff is just kind of doable by whomever. I mean, certainly he's a genius for, you know, being the first to come up with this specific style of orchestration that allows you to kind of run an AI and you know, command it from anywhere. But I mean, that's just kind of interesting, is it not?

15:23

I just think it's a talent Play. And I think they probably want his roadmap for Open Claw to be the roadmap for, for ChatGPT. I mean, this was the big story, you know, last week, was that OpenAI is going to basically merge Codex, ChatGPT and its Atlas browser into this super app and Codex is really going to underpin all of it. And that's also happening with Claude, by the way. You know, Claude Cowork, you know, Anthropic was telling me is going to be basically the front end of the cloud interface. Right.

16:13

It does feel weird that it's two separate tabs.

16:44

Yeah. And like, same thing with Codex. Like, it doesn't probably make sense for it to be a separate app. Like it should probably be the harness that Chat sits on top of. And so they're going to do that and Peter will probably help them do that. And yeah, I think they just want like the power of Open Claw, but for normies. Right. Which they're well positioned to do. But, you know, it's interesting, I think they're having me up to meet with the team because they really do feel how Anthropic has stolen their mojo a bit with the momentum that Claude code has and co work, especially with, you know, our audience, you know, the tech early adopter crowd, this, the Silicon Valley crowd. It does feel like, you know, I've gotten you Claude pilled, obviously. I mean, it does feel like Claude has really taken off and Codex is growing like crazy. It's growing like a weed. But, you know, it doesn't seem to be getting that mind share and that buzz that Chatgpt enjoyed for so long. And I think OpenAI, you know, wants to flex and go like, look, we have all these users and we have, you know, the Clawfather and we're still here, we're still in the race, we're still cool. I think, I think no one questions that they're still in the race. You know, they're huge. They're, they're much bigger than Claude. But I think being cool really matters. Right. So we'll see how cool it is to have sushi in the open AI library. But that's, that's what I'm going to be doing tomorrow.

16:47

Do you have like a little, you know how at the Cheesecake Factory that they have that little note card where you could check off all your favorite cheesecakes you've eaten? Like, have you eaten all 200 cheesecakes? I'm picturing you with like a checklist of all the tech cafeterias around the Entire peninsula. Just checking them off one by one. I know you already had. What did you have? Lamb chops at one of them.

18:02

And Anthropic. Yeah, that would. Honestly, I would do very well there. Stacked against most people. I have been to most headquarters. Actually, that's a good idea. I should just start, like, taking more notes of dictating to my granola. You know what the lunches are at these places.

18:21

I think this is a new revenue stream opportunity for you.

18:38

Yeah, but. Yeah, man, that's. That's my week with that. Should we. Should we kick it to Sam?

18:42

Let's do it.

18:51

All right.

18:51

Support for this episode comes from Anthropic, the team behind Claude. I've been using Claude to help me resolve those messy contradictions at the heart of brand strategy. How do you triangulate one message across a company's products, beliefs, and future vision? Claude doesn't just tell me what I want to hear. It feeds me different strategies across a spectrum of approaches. More like a partner, less like an answering machine. When you need to get up to speed fast on something complicated, you can just point Claude at it and it pulls together sources, surfaces where people disagree, flags the things you should be asking about, all with citations. You can go check. When you connect Claude's Cowork feature to your Gmail and Google Drive, Claude will read your actual emails and docs along with you instead of waiting for you to paste things in. Then Cowork takes it further. You can point it at a folder, describe what you need done, and Claude just handles it. Queue something up before lunch and come back to finished work. Try it yourself at Claude AI Access.

18:59

This episode is brought to you by Indeed. Stop waiting around for the perfect candidate. Instead, use Indeed sponsored Jobs to find the right people with the right skills fast. It's a simple way to make sure your listing is the first candidate. C. According to Indeed data, sponsored jobs have four times more applicants than non sponsored jobs. So go build your dream team today with Indeed. Get a $75 sponsored job credit@ Indeed.com podcast. Terms and conditions apply.

19:54

What's going on, man? Good to see you. Good to see you.

20:26

How you doing?

20:28

You are between two ferns today. For those who are listening in audio only mode.

20:29

Yes, I am.

20:35

Where are you calling in from? Are you in London?

20:36

I mean, it looks very on brand, does it not, Alex?

20:38

We try. Yeah. The Office came before the brand, so the brand looks like it does. Probably partially because of how the Office is.

20:42

I mean, that's what a real brand creative director does when they Come in and they're like, just reflecting stuff that you already do.

20:48

Yeah, yeah.

20:56

And they say, I feel like there's something for you. It's kind of rustic and wooden and plants, and you're like, wow, this creative director really gets me. And then you just turn around and you're like, oh, he's just looking at the aesthetic choices I've already made.

20:57

Yeah.

21:11

Right before you came on Sam, I was telling Ellis that I run granola during every pod we do. Obviously, I run it pretty much constantly, and he's not been doing it, and he says it's because it degrades the audio quality of these conversations. Have you guys heard that from users that when they run it during calls,

21:12

it can do that on Mac os? I think when you start using the system audio, it changes the sound ever so slightly. Yeah. You'll hear it if you have headphones on and you turn granola or anything on. I think even maybe voice memos does it then it changes the sound a little bit. Yeah. Which is annoying. I'd love to be able to fix it, but.

21:30

Well, at least we don't have to run it through one of those programs that we had to do back in the Mac OS days. What were those apps called? You could reroute your audio and all that bs.

21:53

It had a fiery logo. I can't remember.

22:05

Yeah, and there were some great oldies. I feel like OG Mac apps like Handbrake. Alex and I were just talking before you joined about his experience covering the iPhone jailbreak beat. And I feel like. I mean, there's just so many funny hacks over the years, like, oh, the only way to get background notifications on iOS in the early days was claiming that you were actually a VoIP app, like Skype or something like that. And they were just little calls. And I actually feel like you guys were the very first ones that most people probably experienced that just natively recorded system audio. I mean, was that a new macOS feature when you guys took advantage of it or, like, why was that?

22:07

I mean, we did it because we didn't want to do the bot during the call thing, and we wanted to be an app on your desktop for a bunch of other reasons. So, yeah, that was it. We actually avoided doing the system audio for a while because it was just technically a pain in the ass. It was, like, months of an engineer's time fixing all of the edge cases and making it work smoothly. So for the first six months or so, Granola was microphone only. So we'd onboard everyone manually and Explain to them that, sorry, but if you want to use this thing, you got to take your headphones out and let the sound come out of the speakers and then back into the microphone to transcribe. Was a lot of work. I think Apple went and kind of did the system audio API or whatever afterwards, and now it's much easier to do.

22:43

But yeah, demystify this for me. Sam, is Granola actually recording the audio? Because this is a thing that I've had conversations with many people about because, you know, there's the whole thing about two way consent states and all of that. And have you guys found a creative workaround where you're not actually recording? That's what I think is happening.

23:38

By the way, do you consent to being recorded on this podcast? Sam, get that out of the way.

23:58

Well, you're in London. That's. It's probably like, we'll get arrested if we don't get consent from you.

24:04

Yeah, yeah. It would be a good solution to just call every meeting podcast and it'll be implied that it's being recorded. Yeah, I think basically, no, we send the audio to the transcription API and they convert it to transcript and then immediately, Right? Immediately. Yeah, yeah. And they don't save it. And then we also don't save it or. Yeah. So we just. All we save is transcript.

24:08

So I'm not recording people. I can say that legally.

24:32

It depends which lawyer you ask in which state. So, yeah, sure. Nothing is very clean cut. Annoyingly, in this world.

24:36

An idea for prompts in the product based on what state you're in. You all could nudge people and say, make sure you check.

24:45

Yeah, you know what? We could do that. Yeah. Yeah.

24:51

Well, so you guys have some big news worthy of a congratulations. $125 million round, over $1 billion valuation. You're a new unicorn. Congratulations. How does that feel? I know that was on your bucket list, Sam.

24:53

It's like validating. It's nice. Yeah. I mean, the thing I feel in the valuation and that kind of stuff increasing is just the attention we get as a company. And that mostly manifests as for hiring and things like that. Just more people know about you. And especially in London, we get to be like one of the big fish, you know, like, if you want to come work at an exciting startup in London, then we're in a much smaller pool of companies now and we get to kind of pick from the best people. So that I'm excited for. Yeah.

25:14

That's the thing I hear from other founders is that the valuation really matters for hiring and it's one of the main motivators for increasing valuation as fast as you can.

25:44

And for dates as well.

25:53

Right? And for dates. But yeah,

25:54

I'm a married man.

26:00

You guys are opening an office in SF soon, right?

26:04

Yeah, yeah, we actually like five of the basically the whole new San Francisco team like flew over and it's their first day today, so they all arrived this morning jet lagged and having like a rapid onboarding couple of weeks here to meet the team and get a franchise.

26:07

So Elise is secured.

26:23

Yeah, we have the office. Like Ursula, who's made this place beautiful, is over there making that place beautiful at the moment.

26:25

Important question. What other tech companies are you next to in San Francisco?

26:33

We're like a couple blocks from all the other tech companies. I don't think there are actually any next to us. Not that I know of. But we're going to be next to the Moscone center. So like 4th and Folsom, I think or something.

26:38

Uh, just watch what time of day you're walking down the street.

26:50

Uh, yeah, yeah, you gotta, you gotta, you gotta turn right outta the office. Not left. Yeah, I mean to go right. You're on the good side in it. And it's like. Yeah, yeah.

26:53

Well, here's what you do, Sam. Once you decide which company you want to poach from, you find out where their office is and then you just put up a bunch of billboards right outside their office and then that starts the conversation and then you know, you kind of do like an airplane flying overhead that says new Unicorn, join the crew. And then you know, you start the pipeline from there. That's best practices at least.

27:00

Did Alice help you with these billboards that I'm seeing everywhere?

27:23

Jose, Alex's favorite question to ask, did Ellis do anything as part of the project? And at a couple years, a couple weeks ago we had Daniel from Star Boy on, and he's like, actually no, I didn't use any of it.

27:27

Which was great. And you can say that, Sam, this is a safe space. But I do remember Ellis, I think took full credit for Presence as Power as a shower thought in an episode a few weeks ago. Is that true? Did he really come up with that?

27:39

Yeah, I was chatting with Jack, head of product marketing guy, just before this and he said the same and he was like, I was like, was that true? And then apparently yeah. Jack went and looked and Presence is power was the thing. You came up with Alice months ago and I think we, we got to that and then we went On a big idea. Amaze. And then we came back to the presence thing again for the billboards.

27:52

Yeah. And Alex didn't make that sound very nice about taking full credit. I never aim to take full credit. There's always. He's just trying to get me in trouble. But the interesting thing is that I didn't know if that was me or not because I'm just always working on so many stuff, so many things. And so you just go into whatever AI you're using, and you're like, was that me? And it's like, actually, yeah, this time it was.

28:18

Did you not run granola in the shower, Man.

28:41

I think we have good April Fools, though you probably don't have time to prepare with just a week before April Fools. Or do you already have some April Fools?

28:45

No, we have some ideas flown around, but nothing locked in, so we're kind of fine.

28:54

I mean, no pressure. They're mostly bad. They mostly hurt your valuation by 7%. Any company that does an April Fool's joke. So, yeah, no pressure.

28:58

Ellis has AI Butthole icons in our show prep notes, and I just want Ellis to ask about that. What is that about?

29:08

Yeah, well, once again, Alex trying to get me in trouble, which I also deserve because I'm always trying to get him in trouble. Yeah. How did that conversation go down? I mean, obviously it is different than the little Claude butthole I have up in my. Up in my extensions bar right here, or the OpenAI techno butthole icon. But, yeah, I mean, it's certainly adjacent, right?

29:16

It's the whole adjacent. Yeah, I'll give you that. Yeah.

29:40

Surely. Surely you knew that going in. How did that conversation go, Doc? I mean, that's a huge deal to decide what symbol. And I mean, that's what I'm saying. It kind of looks like this piece of wood behind you. It's clearly connected enough to the brand to feel real and authentic. But were there any others in the mix? How did you decide on this one?

29:43

We. I mean, we went through a lot. Like, I think the overarching thing with the. With the brand was, like, we want to make something that feels, like, organic and like. Like a. But like. Like, honestly, the metaphor is kind of stretched on the. On the logo was like anything else. We're going to get into it.

30:01

We're preparing you for the type of, like, gallows humor you will find among the highly ratioed parties in San Francisco.

30:29

Yeah. Yeah. The thing with the brand was to make it feel personal, human. Not like, you know, not like something a corporation would have come up with. And like, so, like, the rough hand drawnness of the thing was, like, was a lot of what drew us to it. You don't intend for these things to come out looking like everybody else's, but then you look at a row of us, Claude OpenAI, and you're like, oh, boy. Yeah, okay. We came up with a similar thing, but we kind of need to feel like we're one of those. We're in those ranks, I think, a little bit. And I don't know, it's a pretty fundamental symbol and therefore I think it feels very recognizable of all of the items and like, all of the options, like, put in a dock next to all the other apps our users are using. Like, this was the one that stood out and felt like the most. Us.

30:40

Well, so that's the right criteria. Is. Does it stand out?

31:32

Yeah. But, like, when I look at my Mac menu bar, I have no idea the difference between any of these apps because the icons all look so similar. I have like six versions of the same squirtle or whatever you want to call it. But I see your alls because I like the way you put the meeting name and how much time is left next to it. That's been super helpful. There's just so much attention to detail in this app. I mean, Sam, if you don't know, I've talked to Chris, your co founder, about this, but I'm like, you're always one of your biggest proponents. I think I've personally converted more people to granola than any product that I've used in the last few years just because I like it so much. But the attention to detail you guys have in kind of everything from the way that the live widget pops up on the phone when you're about to go into a meeting mobile, to the Mac menu bar, to the way that the notifications work. And you can join something and start the transcript at the same time. Like, just a lot of care. And I imagine that's all super intentional. And I'm just, I'm curious. Yeah, like, how that works. Like, is that just something that comes naturally to you or is that something you have to really prioritize?

31:36

You guys, like, invented so many of the things we now think are just kind of common in the AI meeting note space, like even showing the transcript as proof that it's actually working.

32:44

I credit Chris with a lot of this. He's great at being ruthless about what are users actually going to engage with and where are we just getting our hopes up that We've done a cool thing and that it's the might not actually land. I think the main constraint I find helpful working on granola that I've never experienced another product before is meetings are just such a weirdly stressful moment in your day, especially the moments that you click buttons in granola. It's right at the beginning or right at the end of a meeting where you're usually running late. Usually you've forgotten who you're supposed to be talking to or why this thing's

32:55

happening, or you haven't prepped for this podcast like Ellis today.

33:40

Yeah. You're in scrambling mode. Right. Trying to figure out what's going on. What should I do? So you have so little of your brain available to entertain a tool like granola, and so we kind of have to just work with that assumption that you're going to care about basically nothing unless it's going to get you to where you need to go, which is into the meeting. In the right setup, you don't really care about your note taker at the beginning of the meeting. Yeah. So many of the product decisions are kind of downstream of that constraint, I think.

33:43

Could you be a little more proactive, though? Say, like, hey, because you've outsourced your brain to me, you actually need to follow up on this thing that you were going to check the transcript to remember and, like, nudge me a little bit more. I tried that this morning where I said, what do I need to follow up in? And it stressed me out and I immediately looked away from it. But a little more proactive in the moment nudges of, like, hey, this was a thing someone said you need to do by tomorrow. It's now, you know, almost midnight. Something like that.

34:13

I don't know. Yeah, we have a lot of, like, we're a lot of prototypes for that kind of thing. Yeah, there's a lot of nuance, it turns out, which I think we didn't really appreciate until we started trying to build things in this area. Like, there's just a really high bar for, like, suggesting something that you might want to do to, like, where it's, like, actually a thing that you appreciate and not a thing that pisses you off. Like, meetings are so full of people saying, oh, yeah, we should do this thing, or, like, yeah, I'll follow up and we'll talk about it offline or whatever. All these code words for, like, I never want to do, like, think about that thing again.

34:42

I'll have the team look at that and get back to You, Yeah, yeah, you all could decode the bullshit things that people say in meetings when they don't actually want to follow up and then filter that out.

35:17

Yeah, yeah, that might be a good April Fools.

35:28

Like.

35:31

Well, it is a funny thing to associate with, isn't it? Like we're about. I know you say we're for people in back to back meetings and those are typically the people who you could argue hate themselves the most. How do you design for that type of use case? And I mean, what do you feel like? Even just as AI automates a lot of stuff, what are meetings going to be for in a year or two, would you say?

35:34

As long as people are going to have to work together, that's going to be really messy. I don't think there are going to be great ways. AI is not going to solve that. It'll make so much of the rest of our work really efficient, I think. But that thing of humans agreeing on what you're going to do next, I think just always innately going to be hard. And so in my mind, meetings just be more and more about that. I could see a world where an engineering team needs to decide what to do about the thing and then, I don't know, coding agent things are listening into the meeting using granola or something else. And then as soon as the meeting ends, they're going off and drafting up all of the work that they said was going to be done. But people are still going to have to meet and decide on what to do next with that and which of these things should, should go forwards and which ones shouldn't. And I, I don't see that going away anytime soon.

36:02

I saw the CEO of Replit Share after you guys launched the MCP stuff and they built the connector. You know, the idea that an engineer could just have his agent go build something as the meeting was going, as it was listening in on the granola. So that's maybe already kind of technically possible. It's wild.

36:56

Yeah, yeah.

37:15

I mean that's how Alex actually uses this podcast is to make feature requests. And so maybe next time you're on Sam, you'll be able to say by the end of the podcast, oh, we just fixed it, it's already shipping.

37:17

Oh, I have a few more. Just wait. But yeah, no, I mean that's. That is pretty wild to think about introducing agents to the workflow and like you're mid meeting and they're doing stuff based on what is happening.

37:28

Yeah, yeah, yeah. The hard thing there is like you need to will talk about like, oh yeah, granola. Should I want Replit to be able to work on the meeting while I'm having the meeting? And so we can give that snippet to the AI. But the hard bit is knowing all of the background information about you and about the words you use for things and the people you know and the projects you're working on so that we can decode, we can expand that little sentence into a long enough prompt that, that that agent could go off and actually have a shot of building it.

37:40

Talk about this other stuff you guys are launching. You've got spaces, connectors, the API analytics. Seems like you guys are really picking up the pace on. On shipping stuff. Is this just kind of rebuilding Microsoft's suite from first principles or like, what. What are you doing here?

38:14

That's certainly the fad right now.

38:33

I would definitely not trying to do that. I think there's two things. One, we have a lot of teams adopting granola, and that's kind of come initially just from individuals discovering it on the Internet or from you, Alex. And then it spreads organically within the company. People talk about it with each other, but a lot of people have thought about it as basically a single player, individual use product. But we've kind of known internally and within our internal version for a long time has been much more multiplayer. And notes are all put into shared spaces so that you can kind of chat with all of your company's context and know what's happening across the whole company. And it just feels really, really powerful. And so a lot of the announcements coming out this week are about kind of the groundwork to let teams pull all of their meeting context in shared spaces. And the idea being that we'll be able to build a bunch of cool shit on top of that to help you understand what your customers are saying or what projects are going well and which ones aren't going well and what's the. For a higher person at a company, what's the summary of what's going on in this area versus this area, that kind of thing.

38:37

Can you give us an example of how you all have used the shared notes internally to do something versus how it would have been done before?

39:58

We have a daily briefing going around internally at the company, which Granola generates for us every day as going through everything that's happened in the shared granola and picks out the 10 most important things and list those off to you every day. I find that I feel like companies got big enough recently where I don't have tabs on everything. That's happening in every corner of the company anymore. So it's really helpful for me to I see the corners of the company that I wouldn't see what's happening otherwise. There's another one I think is the sales calls folder. We have all of the sales calls pulled in one place and that's fascinating to chat with for understanding. Hey, I'm thinking about building this thing. What have customers said about this before and would you think that would work for them or not? And why it's really useful as a thought partner there.

40:07

How's that sales pitch going these days? I know it's not easy necessarily to make the pitch for paid product that often already exists for free, whether it's inside of Notion or Office or Zoom or otherwise.

41:05

I think we're fortunate that like nine times out of ten those sales conversations happen because there's like a passionate group of people already using the thing inside the company. And so the job of a salesperson at Granola is to kind of like be on their side and help them win over the procurement people at the company. But the argument is like a lot of the argument honestly is like the people like to use this, they'll use this. They probably won't use the built in thing on Google or Zoom or whatever because technically you have it. But also if the experience isn't good and if it's hard for me to use the output of those notes in the different ways I want to use it, then just realistically people aren't going to do it.

41:20

Well, why do you think this feels to people that it needs to be a separate app? It seems like there's something resonating there just as this like home base you've made for all your conversations. And I feel like that's also allowed it to resonate outside of tech as well as opposed to just an add on.

42:07

You're agnostic. It's because you work across all of them. Whereas like Zoom only the Zoom thing only works in Zoom. Meet only works in Meet. Microsoft thing only works in Microsoft. I mean there's a couple that are agnostic too, like notion and OpenAI's. But that's to me is your all's edge is that you work everywhere.

42:23

Yeah, yeah, I think there's a lot of it. Yeah, yeah.

42:40

And the phone thing on mobile, which I've started to use, where you can use it on a phone call is pretty nice and well integrated. I mean I don't think anyone else really has that.

42:42

Yeah, yeah. The guys did such a good job there's like a horrific sequence of steps you've got to go through to set that up because it uses like a VoIP thing. So just like the, the flow that they made to like hold your hand through, that I'm really proud of.

42:50

Like, although there's really no way to have it run while you're mobile on like a meet or a team call. Right. That still is the thing where I'm like, damn, I really. And can you can. Is there a way to design around that? The limitations of the mobile operating system

43:05

or Apple is like, keeps everything too locked down. Like you can't. Yeah, you can't go to the system audio, as far as we can tell. Like we've tried a bunch of things, really hacky things, but nothing has worked so far.

43:21

You got to build a phone, Sam. That's how you're going to get to a 10 billion valuation.

43:31

Yeah, yeah, yeah, yeah. We have Pierre, our first Android engineer, started last week, so we haven't explored this yet. But I'm like, you know, there's a chance that Android is more open and we can make it work on there.

43:35

Oh, granola to Android. Is that some news you're dropping with us?

43:47

I mean, we're like starting to work on it. It's going to be a while before it's up to scratch, but yeah.

43:49

What about Windows? I saw you guys have a wait list for Windows.

43:55

Windows is out. Windows been out for Windows is out.

43:58

Okay.

44:00

Yeah, I think last year, last summer it came out. It's been, it's been like the version we shipped in the summer was like buggy and not great in a bunch of ways. And there's been a lot of stability work.

44:01

I am fascinated by the MCP stuff and the connector stuff. So my setup has changed radically with this where now in Claude, I've got Granola connected. It has all my Google stuff and you know, I use it to prep for this pod, I use it to write stories and accessing my granola transcripts has been the real context unlock. Right. Where I can just say use that interview from yesterday I did in Granola to start building this. Where do you see this going? I mean it seems like you're all philosophy on this is like let be super data portable out, which is kind of against like what you would think the obvious incentives are of building a product like this. You would want people to like not leave the product. But you guys seem to think that that's a good thing. So walk me through that. Like, why is that a good thing? And then where do you see this going?

44:10

How do you convince investors it's a good thing to be open because clearly something's working with your pitch.

44:57

Well, people feel strongly about this because there was that controversy that you guys were locking down data. Right. And not letting it leave. And your co founder Chris had to put a thing out, kind of demystifying this. And you've got the API launch this week. But yeah, talk about all that.

45:03

It was a debate for a long time in the company. Like, I think we've been a little slow to come out with the MCP and the API. And this was because of this debate. Like traditional SaaS wisdom is that you should keep your data locked in the product so that people find it hard to switch off and move out. The thing we have to balance, I think, is I just don't think that philosophy is going to fly in the future now or in the future. Every AI tool is becoming so interconnected and so much of the power of these tools get so much more useful when you can use them with each other. And I think if we don't play ball in that system, then I don't think we're relevant anymore. I think it's essential. And I also think we're a small team and we can only really realistically be good at a couple of things. We have to really pick the things we want to be world class at. And for us, I think that's meetings and there's stuff around meetings and if we do well at that, we'll earn the right to, to be useful at more things. But assuming that we're going to have our focus on meetings and the stuff around that, then I think there's a huge long tail of stuff that people want to do with transcripts that is outside of that realm. And so this is the way to enable all of those workflows.

45:16

That makes a lot of sense. I mean, locking down your data is just one lever to try and win in SaaS world. Right. And I think basically what you're saying is that this element of it is going to be commoditized enough that it's like it's actually hurting us more than helping us to lock it down. And we want to win in other departments like the product design, the agnostic sense of design of it. And I think that totally adds up. I mean, if the product is no good, yeah, you're very happy that there's an export feature. I mean, I remember when I started using like ROAM research a handful of years ago, I was like, oh, this is the coolest thing ever. Backlinking my whole life is about to change. And then it wouldn't sync and the product was never updated and there were just so many, many reasons that it was kludgy and not really any fun. And yeah, it was pretty nice to have the export feature in that, but I think you're right to kind of assume that people demand that or even expect that these days, especially since we're at a point where browser and computer use for AI are at a point where if you work hard enough, you could just get the AI to import or export all your shit for you from various apps or will be soon. And so it's almost like, what's the point?

46:35

I had Claude prepare me for today by looking at my calendar, my granola and my email, right. And I'm now I'm doing these daily prepare things with it. Is that okay that they have like, they use you as kind of the pipe for that data, but then the interface I'm using is Claude. And then like what's to keep them from eventually getting to building like the back end plumbing that you. Which is. I'm not trying to minimize it. I know it's like difficult but like get, you know, OpenAI has it in chat. I don't know how many people use it. I don't know if it's like affected you guys at all. But like what's to keep these like big scaled consumer chatbots from just like, you know, building this. And then they've already got the interface layer with the MCP stuff.

47:49

I mean there's a world where they, they do and they hit it out of the park and we're toast. But like, you know, like my, my belief is like all of that stuff about like meetings being messy and like you're super busy and we've got to, we've got to like design the product that's going to work for that messy reality of you in meetings. It's hard and it takes time to get right and I just believe that we have the team to do a better job at that than a company that's got its attention fragmented in lots of different directions. I also think, I guess you had the time and the headspace to think to go to Claude to get it to prep you for this session. But that's not the case for most meetings. Most meetings you want a similar thing. You want to be smart about who you're meeting and why it's happening and stuff, but also you're ricocheting from one thing to the next and you're not going to have the time to do that. So Granola Inserting the information you need to know at the right moment right before the meeting is a lot of the value. I think.

48:32

Support for this episode comes from Anthropic, the team behind Claude. I've been using Claude to help me resolve those messy contradictions at the heart of brand strategy. How do you triangulate one message across a company's products, beliefs, and future vision? Claude doesn't just tell me what I want to hear. It feeds me different strategies across a spectrum of approaches. More like a partner, less like an answering machine. When you need to get up to speed fast on something complicated, you can just point Claude at it and it pulls together sources, surfaces where people disagree, flags the things you should be asking about, all with citations you can go check. When you connect Claude's Cowork feature to your Gmail and Google Drive, Claude will read your actual emails and docs along with you instead of waiting for you to paste things in. Then Cowork takes it further. You can point it at a folder, describe what you need done, and Claude just handles it. Queue something up before lunch and come back to finished work. Try it yourself at Cloud AI Access.

49:41

This episode is brought to you by Focus Features. Would you let AI pilot your plane? Raise your child? Decide your future?

50:37

On March 27, Focus Features presents the

50:45

AI Doc or How I Became an Apocalyptomist.

50:48

Critics and audience at the Sundance and Southwest Film Festivals call it the most

50:51

urgent movie of our time.

50:56

The AI doc or how I became an apocalyptimist.

50:58

Rated PG13, only in theaters March 27.

51:01

Well, so speaking of like granola, feeling like it knows you. And as you've said in your in your public communications, including for this raise, that the context really is going to be king and is really going to matter. Help you execute a lot of this knowledge work, but how are you able to actually leverage it given the current state of context, windows and whatnot? It's like, do you have a prompt that you're really proud of that allows you to pull just what you need in a useful way? Because I mean, Alex's context is friggin massive at this point. It's thousands of hours of meetings. Obviously it sounds like, oh yeah, it knows you because it has all that context. But it's like in reality it, it does need to somehow turn that into a way to understand and give recommendations. I mean, do you do that in any special way?

51:10

Nothing super special And I don't think there's not anything there that we've figured out that other people haven't figured out. I think honestly I want the product to be way better at this. It's like okay at this so far, but not amazing. And definitely at the 3,000 meeting level, if you ask it a question across all 3,000 of those meetings, it's not going to do a great job. It's like, fine. The average question someone asks is like, hey, summarize what we talked about in the meetings with this one person or find me all the meetings where we talked about breakfast and then write something about that. And with that level of focus, it's great. Granola crunched Spotify wraps thing we did over Christmas did that. And it was like a lot of engineering work to make it do that. And the processing to get everybody's granola crunch spready was this job that took four days of a machine constantly running to get them all.

52:01

That was half the raise was the tokens. It was expensive. Well, do you need better speaker identification? Now we're getting back into product recommendation territory, especially on mobile, where the calendar or when the calendar invitees are not very present or visible or like there's people in the meeting that are not represented in the calendar invites. Like that's the thing I find that also hurts with like the context retrieval and all of that is like granola doesn't actually know who's saying what.

53:03

Yeah, it's the biggest hit to all. Like the quality of the notes and, and everything is like identifying the speakers, I think. Yeah.

53:37

Do you have a solve for that? Can AI fix that?

53:43

We do some processing afterwards to have a shot at it and it works in some cases better than others. We have some prototypes of. There's a few different ways of attacking the problem and honestly I don't think there's a golden bullet because what works on a zoom call where you're all independently on your own laptops might not work on a. An in person meeting on your mobile. So I think we just got to take a couple of different approaches.

53:46

Can they go around the room and say who they are and then it matches it to the voice or something?

54:13

Yeah, this one we tried voice printing. You take a voice print for everyone in the room and then match it to who they are.

54:19

I'm saying this selfishly because I may need to do this tomorrow in an in person meeting where I'm going to be running it on mobile and with six people. So how do I do this?

54:28

Elastic won't Edit. It's not in the product. It won't do it right now. It helped, but still the quality of it was bad enough that it would. Like, there's like, three of us in this call right now. Right. And it would assume the notes would come out and there would be, like, 12 different speakers assigned in the meeting. You know, like, it's okay, but not good enough yet.

54:36

Well, so the fact that this is all you build is also, you know, ostensibly an advantage versus the others who may have the AI part tacked on. Where do you see the most leverage from the product point of view to innovate and come up with new things, since you're the only one, since this is your main focus?

54:57

The stuff I'm most excited about is, like, most meetings kind of create a bunch of work off the back of them, right? And they're often pretty simple, benign things like just like writing an email to somebody saying thank you, or like telling person A that you. That you agreed with person B you were going to do this thing or scheduling the next meeting.

55:13

You're giving me anxiety just saying all this. Yes.

55:36

Yeah. And they suck because they're individually not that hard, but you're busy during the day, and so you don't keep on top of them, and then they stack up into this huge pile of things that you got to do. And it sounds kind of boring, but I'm excited for Granola to really help with all of those. And I think what's interesting is each of those has so much nuance in it to get it to a point where you'd actually use the feature. An email has got to be written in your voice, and it's got to be like, it's got to solve that problem of only talking about the things you actually agreed to do, not the things you just kind of suggested or kicked under the rug.

55:40

Yeah.

56:22

Each of those things has to nail all the social nuances to be actually useful.

56:23

So a Granola email app. Is that what we're getting?

56:28

You might be able to send Granola emails from Granola. We're definitely not building a whole email client, but, yeah, something like that.

56:32

I have ideas. I am having a breakthrough Eureka moment on the pod. I mean, obviously you're starting with the notes, but why not have a moderator who could tell, like, Steve to shut up when he's been talking for 18 minutes straight, you know what I'm saying? Like, so much of the meeting rigor comes from having a project manager or a leader who really knows how to structure meetings properly. And at the moment, you're probably not really attacking that at all. Like, making sure people actually come with what they need to have read, what they need to have prepared. Like, that seems like as big of an opportunity as any. That applies to billions of people, not just people who are in digital meetings. Like, how do we actually make meetings useful and better and so that people aren't hating themselves at the end of. At the end of each one. Whether it's a referee, maybe you could choose the Persona, do that.

56:38

Well, you guys have recipes, right? I mean, that's getting at that idea.

57:37

Yeah, yeah, yeah. And they're like, the, like, personal coaching recipes are by far and away what, like, some of the most used. Like, people, like, love that.

57:40

I was reading that one. Told your. Your co founder Chris to stop acting like a product manager and start acting like a CEO, and he had to go take a walk.

57:50

Yeah, yeah, yeah. I think I've had a version of that, too.

57:58

You've had that experience?

58:01

Yeah, yeah. It's not like, I've never, when we were user testing that, like, you know, we'd, like, get on a video call with a user and let them try it, watch them react to it. I've, like, never worked on a thing where that's got such a strong emotional reaction. Like, you'd visibly see people, like, flinch or like, like, I mean, it's therapy. Yeah.

58:02

For everyone.

58:23

Yeah.

58:24

We should get one to judge our podcasting abilities.

58:25

Alice and say, Alex, based on the micro expressions Ellis is exhibiting, you're really breaking through with some of these critiques and insults.

58:27

What about, like, recipes is unique. Like, what is the thing about them that you think is worth investing in and continuing to. To build that out. I mean, I could see one for, like, say, help me come up with YouTube titles and summaries and stuff for the pod.

58:39

Right.

58:55

Like, which is something we've started doing for the show. But I'm just using, like, standard granola. I should probably use a recipe. Yeah. What motivates someone to make a recipe? And why is this a platform worth investing in?

58:55

A lot of the value is when you have a team of people using granola and there's one person usually that is really motivated to make the team more efficient. The kind of person that likes geeking out on new tools and figuring out all the optimal ways to use it. It's a way for them to go and create a bunch of things, geek out on making stuff in granola, and then for the rest of the team to benefit. I think a lot of the recipes that get the most traction. Are you a recruiting firm and you have this esoteric way of evaluating candidates or the kind of criteria that you look for in the candidates and they let you go and create a recipe that perfectly pulls out all of those talking points from the candidates and present it back to everyone in the way that they like it? Yeah, I think it's the stuff that's super tailored to your company's weird ways of working that hits the best.

59:09

As you are developing the product, I imagine you're getting a lot of feedback from a lot of people, but you guys have some pretty interesting folks on the cap table. You've got Toby from Shopify, Guillermo from Vercel, Kari from Linear. I wonder what have they been telling you? What have they been requesting as. As you go through kind of this inflection point as a company?

1:00:02

Guillermo is a good. Keeps us honest on the quality of the app. Like, we get a lot of UI nets and things like that from him, which is really helpful. A bunch of them have been helpful in just talking to them about how they think about their whole team using it. They're CEOs at the top of very like big, successful, influential companies. And so I was kind of taken aback by how deliberately folks like that are thinking about having a meeting intelligence type system in their company. It's kind of become a requirement that you have a way of capturing the conversations in the company because they kind of all realize that that's going to be such valuable context for all of the AI. Like internal AI agents doing the work in the future. Yeah, just like understanding the language they talk about there, like what they prioritize there, has been really helpful and it's hard to get that insight without having folks like that.

1:00:23

Well, there was a story today about Zuck building an AI CEO agent. I mean, I think they're all thinking about how can they use these tools to help run their companies. Right. And granola is a big, big part of that.

1:01:28

Yeah, yeah, yeah.

1:01:41

Was it you that told me, Sam, that like the truth of the company is in the last meeting, not in the documents?

1:01:42

Yeah, yeah, yeah.

1:01:49

I'm still thinking about that. Can you talk about that?

1:01:50

Yeah, I like it, like continually takes me back, like how much you can understand about what's going on in the company from just the transcripts of the conversation, like basically any other kind of like documentation or like whether it's like notion docs or Slack chats or whatever, they all trail behind the most up to date truth of what's happening. In the company. I think mostly the truth of what's planned or what's going to happen or who's doing what is decided in meetings. It's like people talking about it and then that gets codified in a slack message or someone takes a few hours to write up a notion doc on the thing. And so all of those things are lossy and trailing indicators of what the truth is at the company. Just by having the context of what everybody's talking about. The exhaust themes of that is the truth of what's happening at the company. The downside of it is that somebody sits down to write a notion doc. For example, it's a moment where everybody agrees that this is the thing, we're all agreeing on this thing that we're all writing down. Whereas conversations, people say shit they don't mean all the time. And so picking apart what's actually happening versus what somebody spitballing at lunch is not trivial. But yeah.

1:01:53

And have you said how many users you have at the moment? I know in the new release you talked about some companies like some big ones like DoorDash that are using you guys. Where does that all stand?

1:03:13

Yeah, we don't talk about absolute users numbers publicly, but DoorDash, Brex, Vercel, Vanta, a lot of the big, most AI forward companies that are adopting us have adopted us or we're talking with them at the moment. I think we just had so much organic viral growth from person to person, especially in the Silicon Valley ecosystem that's kind of bearing fruit. We've hit critical mass in enough of their companies now where it's turning into sales conversations. So the sales team have their work cut out like keeping up with that.

1:03:23

But I got the notification about a month ago that I had 30 days to pay or lose access to everything past 30 days.

1:03:59

It's been a while coming.

1:04:09

Yeah, I was waiting. You know, we were in the like, you know, free Uber.

1:04:10

You weren't paying?

1:04:14

No, I wasn't paying this whole time. And I was like, man, did I get a hookup? Did Chris hook me up? Because he, he helped me with an account transfer before you guys let your let people switch emails. And I was like, damn, maybe I'm on some great white list or something. And then I got that and I was like, nope. And then I was also thinking, well, like maybe we were in the subsidized Uber phase of this.

1:04:15

Yeah.

1:04:36

Yeah. And I imagine now the revenue ramps looking a little different now that, that, that, that time has come. Was that always the plan? Just like we got to, like, you know, I think. I think saying, you know, you only get 30 days worth of meetings is. Is makes sense. I know it's probably expensive to store that stuff at scale, but, yeah, it's quite a move.

1:04:36

And we had to start making money at some point.

1:04:58

This is AI. What are you talking about? You don't have to start making money.

1:05:02

I don't know. We kicked the can down the road on actually charging for the product for a really long time. I think when we launched, you could technically pay for granola and we had pricing on their website, but really just the priority was just make the product good. And we were a tiny, tiny team. And building a paywall would have meant stopping everything else. So it took us a while to actually get round to doing it. There was a point where the user numbers started getting big enough that the amount of thing was costing us. You could plot a chart and the numbers got really scary big really quickly. So, yeah, that was the motivator.

1:05:05

I know the conventional wisdom was always users first. And I think part of that is that you prioritize the users first for a number of reasons, but one of them is that they start building their data set within your product, and then it's sticky and or hard to leave. But now you kind of look at Twitter just kind of coming back to the defensibility point of view, and there's another Steve out there who vibe coded his own granola and like, one day or whatever, and that is new. And then, you know, there's also the narratives about the SAS apocalypse of, like, capable engineers at doordash just coding it for themselves. How are you guys thinking about that? Does that keep you up at night?

1:05:44

No. Maybe it should, but I think it's like, today, at least, it's still the case. I think that it's easy to make something that looks like it does the job, but to make something actually good enough that you rely on it, you've really, really got to trust granola to be doing his thing and to not. To not let you down because you want to hit the button and then put it out of your mind. And, yeah, if we fail 1% of the time, that's just not okay. Users will churn and stop using us. And doing that well enough for a large number of users is still a meaty problem that I think is safe for a little bit.

1:06:24

I guess, when it comes to both the reliability and also the data security, it's like, yeah, you do want a company you could trust as opposed to Steve's app that just happens to be free. Right. And so maybe categorically, you're in a better place than most who. Who may be called rappers or otherwise. Is that still a term that's being used, or is everybody past that?

1:07:02

Yeah, man, I haven't heard it in a while.

1:07:23

OpenAI is just an Nvidia wrapper, so you can just wrapper this all the way up.

1:07:26

Geez, that was like the number one raising the first rounds. That was like the number one thing we had to get over was. Was like us just being a rapper and times have changed.

1:07:31

Sam, I noticed you guys are hiring for a community manager in San Francisco, and one of the job description points is to make granola's SF office synonymous with the forefront of tech in the city. What are your ideas for doing that? How are you going to do that?

1:07:41

We throw a lot of events here in London. Meetups, firesides, hackathons, design evenings, all kinds of things. It's been so good for us. One, it's just really fun. Everyone on the team loves being able to just go downstairs on a Wednesday evening and hang out with new people and go see a talk by someone interesting. But for just as a small startup, I think we're just known at this point in the city. Most engineers have heard of us, probably buy our events more than our product. So for hiring and for attracting interesting people, it's been like, really, really good. So, yeah, we have to figure out what the San Francisco version of that is like. I think there's a little more competition for tech events there, so we might have to be more creative.

1:07:58

There's a lot of competition. You may have to bring in some, like, sushi boats or something. I mean, you're. I mean, it helps that you're. You're not next to something, you know, if you were next to a competitor, like literally next door, you know, but you've also got that. That Moscone area to contend with and people getting there.

1:08:43

Yeah, I think we're going to make the office really nice as well. Like a place people, like, just have to go see for themselves.

1:08:59

Yeah. What's going to be the bougiest thing in this office? Are we doing, you know, haircuts? Are we doing laundry?

1:09:05

Yo, yo, yo.

1:09:11

Obviously we need fancy, artisanal granola, which can be banging if you've had the good ones.

1:09:12

We have granola companies send us their granola. You know, like, they're like, oh, these people must need good granola. We'll send them a trial package. So we, we usually have like at any given time there's like a handful of different granolas for the people who sent in to try out.

1:09:21

Is there a most popular granola in a terrible granola that's actually all sugar? Is there a slack channel around that I assume. I hope there is.

1:09:36

We have a chief snack officer and a chief drinks officer and I think a condiments one as well. And they are getting. This is a busy job. They're getting things from all directions on with feedback on these things. Yeah, yeah. Although I think we have an OpenCloud bot running in the office at the moment and a lot of their job is collecting all the feedback from on the snacks and the granola and stuff.

1:09:47

How are you feeling with the limited time we have left? How are you feeling about London's tech scene overall? Is it up? Is it down? Is it.

1:10:10

It feels up. It feels up. Like when we started, what, like three, three years ago now, it was super research heavy. Like I guess DeepMind felt like the place that everybody talked about as the happening place in AI. And there's more diversity now. I think there's just more companies doing more interesting things and there's a wave of AI startups that I guess started around the time we did and are now bigger and have their own centers of gravity. So it's looking better. It's still, I think tiny compared to San Francisco, but. But we have good people here. Yeah.

1:10:17

Nice. Well, we'll let you get back to your snacks. Thank you for joining us and yeah, congrats again on the big raise. We'll, we'll keep in touch.

1:10:52

Yeah, let us know. Thank you. Thanks for coming on Sam. That's it for this week's show. Please don't forget to like subscribe wherever you get podcasts. We are access show on the Interwebs where you can get us an all formats. You can find us in video. Please do that at access pod on

1:11:00

YouTube and if you like this episode, please leave a 5 star review or thumbs up and share with a friend or agent.

1:11:21

You can find my newsletter at Sources News.

1:11:28

You could find me at Hamburger on twitter and@meaning.company for all your startup storytelling needs.

1:11:31

Access is part of the Vox Media Podcast network and this show is produced by Hooked Creators.

1:11:36

Bye bye. Thanks again to Anthropic, the team behind Claude for supporting this episode. Anthropic is committed to keeping Claude ad free. Your conversations in there won't be shaped by advertisers which given all the details of your life you share with AI is worth paying attention to. And for the people who want to dig deep into something, whether that's writing, research or just working through a hard problem, Claude is one worth a shot. See why problem solvers choose Claude as their thinking partner? Try it yourself at Claude AI Access

1:11:41

New Year, New Me Cute, but how about New Year, New Money? With Experian, you can actually take control of your finances. Check your FICO score, find ways to save and get matched with credit card offers, giving you time to power through those New Year's goals you know you're going to crush. Start the year off right. Download the Experian app Based on FICO Score 8 model offers an approval not guaranteed. Eligibility requirements and terms apply subject to credit check, which may impact your credit scores. Offers not available in all states. See experian.com for details. Experian.

1:12:15