SUMMIT STAGE
Grant Lee of Gamma discusses scaling from a PLG startup to a $100M company. The team uses "gambassadors" to guide features, blends seat-based and consumption pricing, and hires slow for culture fit. He advises founders to stay purpose-driven and build self-reinforcing flywheels.
Chapters
Ciao!
Two technical teams pointing fingers. The root cause was buried and a manual fix was impossible. Both teams were stuck.
Welcome back to day 3 of SASTR Annual 2026. Please take your seats, our program will begin shortly.
Hello, Saastr. Welcome to day 3. Alrighty. Thank you for coming. I know it's still, it still seems we've been having this debate all week, like 9 and 10, like seems too early at a conference, but like usually I'm on Zoom at like 6:00. Yeah, same, same.
But like physically getting anywhere is always a struggle before a certain time in the morning.
I'm very excited. We've been trying to get Sarah Kennedy, um, who I— can I call you SK on this?
Yes, please.
SK, to a physical Sastre for a while now, so we're super stoked to have her here. I know she's done some stuff with Jason and I on like the podcast, and she's been a big proponent of the Sastre community. So stoked to actually have you here In person. Thank you for coming.
I'm thrilled to be here. It's kind of coming home for me being at Marketo years ago. Yeah, I know probably many of you that I'll get to see after, I hope.
Well, fun. I know you have— we want to talk about Boost Bytes. We want to talk about what you're doing internally. I know you've got a video to kind of set the tone for folks, but I know, you know, obviously Gemini has just taken off in the last year. I was just thinking like between the last SaaStr to now, like You guys have not only skyrocketed in terms of growth, but you're like number one now. So if you don't mind sharing with folks, I know it's maybe not a loaded question, but a big question to answer. Like, how did you guys do that in such a short amount of time? Gosh.
Well, here's what I'll tell you. It took a whole lot of just passion and commitment, I think, from a company that's reinvented itself many, many times over and over again, but I also I think with Gemini, it's been a really exciting ride. And I think when we think about the model itself, like models are constantly evolving and changing and it's happening on a monthly basis, on a weekly basis in some cases. And that's really exciting. But when we look at what is most important, it's actually building a platform that people can build on. And so like we've been really focused on at Cloud for years, our model has always been around being a multi-cloud provider. Provider. We came into the game when others were well established and we were like, okay, we want to create choice and openness and create an opportunity for people to choose how they want to build, what they want to build, the way they want to build, and make a platform that's the most like amenable to that. And so we've applied that same philosophy as we've been building our new agentic capabilities and our agent platform. Whatever model you want to use, you can use, and whatever dataset you want to bring to the party or want our capabilities to apply to, we make that open and accessible as well. So, but it's about making sure we've got the right flexibility in the infrastructure behind it. And the full stack that Google brings is what's really unique about our offering. So that's been honestly the pillar of our success has actually been what underpins everything and not necessarily the new feature or flashy launch. Like that's important. but it's the full stack that actually is where our really just core of our foundational investment and focus is.
And Gemini is really fast. Like, one of the things we've noticed, right? So we hooked up, for those who maybe don't know or didn't apply for, like, the pitch comp that we're doing here today with you guys, like, so we had startups who are coming to SaaStr apply for the pitch competition, and we hooked up Gemini to it, which first I thought I was going to need help to do, but then it was actually really easy. I was like, wait, I got this. So it was actually really easy, which I found interesting. But then what I found interesting was how hard of a grader the model was. It was like, OK, we're hooking up Gemini to our pitch comp. So you put in your basic details of, here's— I'm a startup doing XYZ. I have X amount of funding. Here are my top 3 customers. And then Gemini would give you a score so that we, as the humans, would know who should be the finalists for today's pitch comp, financially.
Right.
Right.
And so I think it's so interesting how consistent and like almost fair Gemini is. I feel like a lot of the other models are very gold-seeking to the humans, right, where they want to please you. They want to give you— they want to give all your speakers a 99 out of 100. And Gemini out the gate was super fast to hook up, super like, again, fair and even. And from your perspective on the, on the marketing side now running this internally, how have you seen that play out where like Gemini has just become almost like a grounding model?
Yeah, I think it's been exciting actually to be sitting at the heart and soul of what we're building and what we're bringing to the world. And being customer zero is at the center of everything. So I think even 3 weeks ago at our big event, Google Cloud Next, he started the conference by saying like, we are customer zero of our own technologies and shared just some of the examples as to how we're doing that across HR, finance, service operations, as well as marketing, which we love seeing that example. But it's really exciting because that lets us both invest in what we need. Like we're able to tell the DeepMind team, hey, this is what's working, this isn't, this is how we— these are the use cases that we think about. This is what we need the model itself to do versus what the harness around it needs to be capable of doing versus what the tools and the applications or the platform itself needs to do. And so we want to be the very best user of our products because we want our customers to be able to be the very best users of them. And so when I sit in rooms like this, it's really awesome to get feedback because that's like hearing from you even and how you're using it this week is incredible. But that's the whole intent behind that, I think. Um, but what we've seen, there's just a lot of innovation happening at such a rapid pace, you all know, and Our teams daily are learning new things. They're sharing. Yep. And we're creating communities around that for them to share, um, and grow. But specifically, like just yesterday, we were, um, we've got a new event we're running and I was just like, okay, why does the sales team not have these contacts? I'm like, it's just like, are they, is it, is it hygiene? Is it that they really don't have the connections? I know they've just met with 3 of these people in the last month. All this was going on and we're like, Why is it not in Salesforce? What's the deal? And so we've got— there's some new capabilities in our kind of martech stack that I was like, okay, why don't we go run this on and take our agent that we've run on some other things and actually apply it and say, all the people that you've met with in the last year to 2 years that are not in your contact list, that are also not in Vector, is our version of Salesforce. I'm like, let's run that. Let's actually pull and source that through. And I did it with just the VPs in the room. Room that I was with at a sales offsite this week. And hello, I was like, there's like little— but those are little bitty, those are little bitty tasks, you know? And so then if you think about the full workflow of what our teams are starting to reinvent, and so task versus workflow is a big part of that for us. And we're trying to take a look at how do we take specific workflows and reinvent those end to end and apply our agent capabilities and what Gemini can do on top of that.
So you brought up a few things I want to dig into. For folks because I feel like what I've seen this week is everyone's at like, everyone's hungry for this information, right? Like people want to get this done with agents. They want to embed it into their workflows. And I feel like that's been the biggest takeaway. So you were just asking me to like, yeah, what's the most surprising thing? I'm like, dude, people are hungry for this info. Yeah, they want to get the agents to work. They want to integrate Gemini. They want to use things like AgentForce. Like people are just hungry to get their agents up and running so that the humans are maybe becoming the manager of the agents or more freed up to do, like, again, higher value activities that we previously all had to do across our org ourselves and within our sales and marketing teams previously, right? But now we can do that with agents. So a couple of things you brought up I want to dig into. The first being one thing that Elena from Lovable brought up yesterday, and I want to hear the Google version of, was she was talking about how because Lovable is vibecoding, and they can just ship things lightning fast. So they have a Slack channel where anybody can vibecode an idea, and if somebody else at the org +1s it, then they build it into the product. You guys have a huge org. I don't think you can quite have a Slack channel in the same way they do. But how are you managing people now having their own agents at Google using Gemini, being customer zero? How do you guys work that back in then to Gemini and all the models?
Yeah, so we actually We have chat spaces that are in Google Workspace and there are thousands of people that are in it. Like you could see for Gemini Enterprise, we have our own dogfood testers internally that are constantly providing feedback and our marketers sit at the center of that. It's really exciting because we're building just like daily, we're building new things, we're learning new things, and our product team takes that feedback and is implementing it. So that's really exciting to sit at the center of that. But I would say, you know, taking a step back, there's a lot, a lot of opportunity for us to teach people what to do. And I think if you look at the greatest friction in a workflow, that is what will be the biggest inhibitor to adoption, well over agent quality any day. And I think us taking a look at how behavioral change and like the change management aspect of it in the training, that's really critical to everything we do. And we've invested a ton in training. And I think even just our data shows inside of just the cloud, the Google Cloud marketing team specifically, that we have the top 20% of adopters are also the ones that are, that are finding the best productivity and they're the ones that have also done the most training and learning. And we started that with something called AI Boost Bytes. Yep. That's been a big part of that foundation for for us, and it's like 5 to 7 minute videos that some of them even 2 to 3 minutes, and it's just like one specific thing that you can do. In the early days, it was like, how do you create slides with Gemini? How are you doing different things that are pretty basic? And now it's evolved into being like multi-agent workflows and how do you do multi-agent orchestration across a campaign. And so we continue to roll these out and I'll show you just like the beginning of one. It's like these are things that we publish on a recurring basis.
Welcome to AI Boost Bytes, where we teach you the skills to supercharge your work with AI. Have you ever been caught in the middle of a high-stakes customer crisis, trapped in a technical deadlock where every team says we can't solve it? I have. A major client was facing a catastrophic failure putting a massive amount of business at risk. Today I'm going to show you how I use Gemini to break that deadlock and drive the solution all without leaving my inbox. Let's set the scene. One of our largest clients was experiencing a mass failure of their Google Ads location assets. The email thread was over 20 messages long with 2 technical teams pointing fingers. The root cause was buried and a manual fix was impossible. Impossible. Both teams were stuck. Instead of copying and pasting this entire mess into another app, I decided to ask my AI partner for help right here in Gmail. I opened the Gemini side panel, which can read the context of the entire email thread I have open. I didn't just ask for a summary, I—
so that we go through that in 20, 30, 40 different little iterations. Right? And so just to give you a sense for that, and that started internal and it was something that got so rapidly adopted. We were like, we had, we were actually having conversations with a lot of our customers about what are you doing to train, et cetera. And we were sharing, we do this and we're like, why don't we just make this available to everyone? So we put it out, um, for all of our customers. It's free. You guys can go and check it out too. But, um, it's really helpful and it's just like really practical things. And then we were having competitions internally. Where it's like you actually did have to go complete the task and show and prove that you could do it. And then, you know, earning badges, which is obviously like gamification from 20 years ago, but like still we're all like it.
Yeah.
I mean, everybody loves a badge. So anyway, it helped us really start to connect with what was resonating and how training was happening while people did not have any time in their day. They really only did have 5 minutes a week, and time is the biggest limiting factor to adoption right now as well. As time to actually spend on learning. So this has really helped us and had that take off inside of our business as well as with our customers.
Yeah, one thing I want to ask about the Boost Bytes was clearly, like, you know, not only are you getting views across the internal order, but now externally, because now you're just sharing this with everyone on YouTube. I mean, half a million views is pretty impressive because I know, um, as you're saying, they're getting more complicated too, right? So people are getting further along in their agentic journey. What are some of the top, like, use cases you've seen internally for Gemini?
Oh gosh, so many. Um, I think in marketing right now we are really excited because number one, embedding it into people's day-to-day workflows is like in the surfaces where you're already working. It's one of the most important things. So anything happening inside of Google Workspace in particular, just the— some of the basic things of just, yes, it's kind of point solution, but it is important work when you're just doing content creation and production that's happening all the time. And it's just really having a thought partner right along our marketers and the work they do. It's been great. On a larger scale across all of Google Marketing, scaled content production is one of the most important use cases that we can both use to save time. It's also a place where the volume exceeds the human judgment required in some cases, which is somehow how— some of the time how we look at that is, Where there's high, high volume and you can apply limited judgment to get a really high quality outcome, that's where we go hard. And that's for us scaled asset production, both on creative assets as well as emails and our product-led growth motion. How do we start to automate more and more of that engagement in those interactions and to do it at a personalization level of almost one-to-one? That's been a huge opportunity for us that we've been able to capitalize on. I would say other things are just like leveraging a lot of our Gen Media tools like Veo as well as Imagine. That's been really at the center of a lot of our work and a lot of how we storyboard. Even actually, and I saw some of the stuff about how y'all have used AI to help plan this event.
We did a lot of it.
It's incredible. But we did same thing on a like with different types of applications of it, but it's been really cool to see our production team and our broadcast team in particular leverage a lot of that capability across all of our events.
I think it's interesting because I feel like of the folks I've asked who do events or want to start to do events, even Loveable was like, we have to do more customer events, more user events because we're getting asked all the time. Replit's the same thing. They're doing their first, I think their first customer event in New York in like a month or two, which is crazy because they did Sasterfest. Now they're doing their own customer event. And they're like, wait, what did you guys use your agents to do on the event side? Because it's such a weird thing for events in particular because it always falls under marketing, right?
Right.
And it's one of those things where it's such a physical aspect of marketing, like these building these screens, the agent could not do. Right. Okay. Right. Folks were asking us this week, okay, how much did the agents or other people Well, people's agents help do the content. I'm like, dude, humans did a lot of the content. That is something, there's still that boundary where agents are great and they helped us a lot for a lot of this, but there are still a lot of physical aspects that the agents can't do. One of the very first things I wanted to try to identify in marketing, 'cause it's such an annoying thing that we have to do this, is work on all the event CADs. I'm like, this is such a boring thing and such a nuanced, I was like, Why do I have to know how to read a CAD? Like, can an agent do this for me? But I asked a bunch of agents and it couldn't get it right. It doesn't, it doesn't understand it yet. And so that was slightly disappointing. But I think that, you know, the agents will get better and eventually it will.
Yeah.
But, you know, for planning things like events, content creation, as you brought up, is one of the best use cases. We used it more so on the promotion side, right? Yeah. So we use Higgsfield a lot, which has Veo built into it. I like that one the best in Higgsfield. I was like, I get the best outfits when I, when I use the Veo in Higgsfield. If you guys missed it, Higgsfield was here yesterday, but it's got all of a bunch of different video generation models. But I like to use the Veo in it because I find it to be the best outputs. But it's, again, it can help with things like that. But it's, it's so funny when it comes to like the physical world where it kind of, it starts to break down almost a little bit.
Right. Well, it's Actually, it's where it comes to life most. I think it's actually really exciting 'cause I, so just thinking about, there's a laundry list of things we use it for at Next, but a few, like, we do onsite customer, like, story capture. So we've got a video studio and it's like, how do you do 100 different personalized interview scripts? You know, just basic things where you can have a lot of human input, but it can help produce it at scale. Yeah. And we have a lot of other, parts of Next. Like we even vibe-coded a couple of things for a demo that's a little lower risk, right? Like you can do some cool things to activate even just some of the animations and the countdown clock that happens before the keynote. But I actually have an example of another, a big project that we did at Next that was really cool. Maybe not so cool for the team whenever I asked for this like 3 weeks before, but everybody—
But agents can move pretty fast.
Agents can move pretty fast, but you know, You know, something.
Yeah.
But, but the interesting thing is like, how can it supercharge like the, the creativity of the humans that are involved in putting on these experiences that are deeply human experiences? There's so much more emotion of showing up and connecting with the community. Exactly. Um, but we, we actually, it was like 3 weeks before the event we had decided, or I decided, I saw, I was looking, we were in the rehearsal stage and I'm like, this opening video that we had put together is just like not hitting it for me. Yeah. And it wasn't, using— it was using AI, but like not enough of it to really feel like we were showcasing our capabilities. So we had challenged the team to go back and actually look at how do you come up with a story and also a narrative that is deeply rooted in our product. And the product is the video itself. But I was going to show y'all— let me see if I've got it in here. Oh, I skipped. You don't need to see the full story next. Here we go. And Weezer was performing at our event, and so for those that are familiar with the Blue Album, you'll recognize the backdrop to this. This is the opening video for Google Cloud Next. I'll talk on top of this if it's okay. So this is all generated by VEO, Nano Banana, as well as some other products that we pulled into the mix, a couple of custom agents that were built on top of Gemini Enterprise Agent platform. We had to do everything from create, I think there were like 138 different Easter eggs in here, but you can see, but actually Taking inspiration from the 1998 garage, Susan's garage, that was the origin of Google, and like prompting our tools to say, okay, how do you tell a story that's also about like, we're trying to build the platform that everybody else is building in their own garage on.
Right.
And so looking at how do you bring those things together, and like some of the things that were produced as a part of this, I'll show you in a second. Like we didn't prompt it to do a couple things like this.
Yeah.
So the water that just shot out, we did— that was nowhere in the prompt, sort of came up with it on its own. We were like, create all these balls and like represent our customers and do it in a really cool way. But it was injecting kind of its own creativity. And it was a really, really special moment to be able to say human ingenuity was intersecting with AI to create something really, really special in this physical environment that produces goosebumps. And I think that's a really interesting thing is where AI and humans are coming together to create work that wasn't possible before. This was not possible in 3 weeks in any other scenario. And like drawn on a napkin, brought to life with Google was how we were like, that's a way more cool way to start the show than anything else we would have come up with outside of that. But it literally did start on a napkin. We had our creative director like sketch this out and fed this into originally Nano Banana and started to produce some of the images and source material and some of the inspo and also said like, go pull from Google lore of the past and some of the Easter eggs like Dino and like all of the other things. The original server rack was actually built inside of Legos. And so having that roll up in the car car that was, I think it was Larry's car that was like the trunk of it, it pops out in the video. There's just like random stuff that's just so fun to be able to pull in, but some of the creativity that came to life just by everything that we aggregated and put into the tool was really, really, I think a special way to show that. And then Nanobanana just producing what it does. If you've used Nanobanana, you know like some of the coolest stuff comes out of what you put in, and so we were trying to figure out what is what does everyone's garage look like? And then introducing motion using Veo. So that was a really special part of it where it went wild on some things. Like some examples here of things we definitely didn't use that look really hokey.
But you're like, the one in the middle.
Yeah, but like they look a little— and so we had to continue to prompt and reprompt. And then when you start to blend sources and take that material and put it together to create something even more And I was like, how do I QC every single, like, line of code? I was like, this is like a barcode. This isn't actually like, is it binary or what? I'm like, how do I, how do I go red team this video? But then like looking at the screen that this goes on and I appreciate like not everybody's running an event with this large of a screen, but it's a really interesting challenge because we also needed to up-res. So going from 4K, which was the max output, and then taking it to 12K, that's a big challenge for all events. Is like, how do you scale up?
100%.
And so we used a custom model from DeepMind to actually do that. It's the same one we used at the Sphere for the Wizard of Oz when we were producing that event. And it was great. I mean, but it was also— we were on a crunch because that model isn't— it isn't instant.
It's like—
so it takes time to generate and regenerate and make sure we can up-res to actually put it in a room that's as large as a 747 jet, or 737.
It's funny, we were saying last year, so last year we used Higgsfield, which is the video, and then like Nana Banana, and then we used Re:ve for like a few different things. Now I've used Gamma for some stuff too. I have a mix of ones that I like to piece together, 'cause they all output a little bit different, but it's funny, 'cause we were saying last year, like we couldn't do things that you just showed, because of the up-res problem, again, very physical problem of like, okay, it looks good on a screen, but then you go to maybe send it to like a customer, and they're like, okay, is it too big, is it too small? And when you have big LED screens, like this one's pretty big, and then the one in the other hangars are bigger, and then yours was an airplane-sized hangar, it's not at the same level of output and resolution that you need for this very physical thing that you're doing. So last year, we tried to do some of this stuff, and we were like, it's just too early. But now, a year later, you're like, dude, it's frickin' crushing it. Crushing it at getting not only all the inputs right to create you this video, which like again, the previous world would be you probably had to ask an agency if you wanted to do something like this with more than 3 weeks, more budget than just using your own internal tools.
Right, and like some of it still isn't even possible, like some of even the creation of all of the bubbles and the way that they moved, 'cause I was looking at like if it was possible to do without AI, like, then why are we doing that if we're showcasing the product in a way? And so we, we tried to look at it like that as well. But like, each of these individual stickers are kind of their own generated thing. And so piecing those things together and doing all of that, it's just like at human scale, it's almost impossible, right? But you have to have very talented creative directors who can know what to prompt. And so we actually, we have a blog about it with like a lot of the the exact prompts we used to get to this video. But anyway, it was a really cool project to go through to know we couldn't have done it in 3 weeks without these capabilities, and we couldn't have ever done that even in 1 year prior based on where things were. Yeah.
So content creation has obviously scaled up a lot in the last year. This is one of those things that's moved really quickly in marketing.
Totally.
Is obviously like content creation. Is there another big top of mind use case? You just mentioned emails for your event.
Yeah. We had a campaign with Chrome. We were launching Gemini in Chrome. That's another one where we were doing scaled asset production. So there were thousands of different creative assets that we were needing to produce for that. So that's one where— and Sundar actually talked about it a little bit at Next, but that was one— that was actually when we were at 2.5 with Gemini even. And that was for us a really pivotal moment to be able to create all the things that we needed to truly personalize at scale and Saved us about 70% of the production time. It went from many, many, many weeks to days. And people say that a lot, but it's true. And that, and it's like really exciting to see that happen. The, the more important thing though is actually that the conversion rate lifted. And so, oh wow. When we see conversion lift, when we see not just volume and output being capable increasing, oftentimes that means quality goes down and quality actually went up as well. And so that's, that's really exciting when we get into the mode of thinking about personalization down to the one and being able to do that at that scale that was never before possible.
And Google, both from a customer standpoint and an internal standpoint, is at such a large scale, right? For some of the startups in the room, where do you feel like maybe they'll get the most benefit out of an Agentic use case in marketing?
Yeah, well, I think right now, I mean, I actually think that's not necessarily different. Google isn't like— We, we as a marketing team are facing all the same challenges that other companies are. We do have deep access, yes, to technologies just because we're, we're also building a lot of the things that we're bringing to the world. Um, but the behavioral change and the workflow change is actually one of the more challenging elements of all of this. Mm-hmm. So I think it's, I, I would, my advice, whatever that's worth, is focusing, picking one workflow, one specific workflow, and not just one not like create a PDF or generate an email, but actually an end-to-end workflow, whether that's like your lead generation workflow or anything that's tied to a campaign brief or campaign creation like that. And taking things that have been more human-centric historically and figuring out how does an agent apply to that step by step. I think figuring that out and doing that a couple of times a year can put you well ahead of where your competition is in many cases. Ways. And I think that's where I would start, is just by focusing small. Focus on, by the way, where you have the highest quality data and the highest pain points. Like, that's a place where a lot of innovation is more, um, just amenable to more on top of it in terms of applying agents.
One thing I want to ask, I also asked Lovable and then the CMO of, uh, Snowflake yesterday, was how have you— how has this impacted hiring? Because the biggest thing I hear is that We want to hire folks that are comfortable with agentic marketing, but they don't really exist yet. Like, it's only been a year really of folks out in the workflow doing things like this with their agents. So how are you tackling that?
So it is a tell me what you built conversation out of the gate, and I want to hear from people about what they're— and again, vibe coding, a lot of people have different opinions about it, but it tells me a whole lot about who you are and how curious you are as a person. Like, curiosity is what I'm hiring for now. And I get into those conversations and you can tell the energy shifts whenever someone's telling you about something they've personally built.
Exactly.
Yeah, it's just, it's different and you can tell if it's authentic. And so I, I get into that conversation a lot earlier in the interviews that I'm having and I'm looking at talent. And I do think like, we'll see how it evolves, you know, the portability of agents, your ability to bring things with you to work. You know, a lot of people have talked about this where your resume is now a collection of agents you've built and what you are bringing with you. It's not just you as a human, it's you as a team. And that being an interesting dynamic. Now, how that technically works and how the portability of that technology will apply to your— whatever you work for is yet to be seen. But I do think that it says a lot about the talent you're looking at if they're already thinking that way, because that's the kind of person you need to build the team that you're in particular gonna need that's gonna be mostly agent-led. So that's how I think about it.
And how internally now do you structure the humans managing the agents? 'Cause everyone just has their own multiple agents now.
Yeah, I heard, by the way, you're looking at having agents, having humans work for the agent, which is interesting.
I know, yeah, so like we are hiring that. We'll see what happens. A few folks have applied and they're like, Who's the interview with? Is it with the agent or the human? I'm like, well, it's going to be with a human, but like, eventually you are going to have to talk to like Tenke, who's our AI VP of marketing, and we're going to see what kind of questions it gives you. Yeah, but it's just because, like, I'm sure you have this problem on a very large scale, which is we can't keep up with our agents.
Totally.
Like, so I'm like, okay, we have Tenke who's doing 3 marketing ideas a day. I can't keep up with 3 marketing. 50 ideas a day with everything else I have to do. So then we almost said it as a joke on the podcast, we were like, well, what if we have now a human who reports to our agents to actually, well, at first, we think this person will implement the campaigns that our agent is coming up with because they're grounded in data. The thing is like our agent 10K who's doing marketing, I think we said he's like 6 months old or whatever. I mean, 6 months as a marketer, right? Running autonomous email campaigns, writing the email copy, plugging into Salesforce, looking across all of go-to-market. Like, that's better than most humans in 6 months. 6 months is usually like training, training.
Yep.
So we were like, okay, I think we're at that point now where we need somebody who we want to implement the ideas 10K has at the start, but then also work with the agent, right? Do you manage the agents? Does the agent manage you? At the end of the day, does it really matter where the lines are?
Like, it's kind of going to be fluid.
It's going to be fluid. No, a related thing I wanted to ask your take on, so I'm glad you brought that up, was somebody was asking me like, where do you think the lines will be now between sales and marketing? Because so many of the agents blur the line of just general go-to-market. So I want to get your perspective on that.
I think that the lines in general are blurring already. And I think they always have, by the way, they kind of were for sure, like for the last decade. So for anybody who's worked also in just like the martech space, we were already there. Like RevOps was the beginning of that.
Yep.
Anyway, exactly. I actually was writing a brief on this yesterday, just, and it's an unpopular opinion maybe, but I just was wanting to start to provoke us around how do we think about marketing marketing's value inside of an organization and becoming closer to sales in a way that is really, really all about accountability. And I don't think the structure doesn't matter as much as it does how you think about your responsibility and your accountability to the work and the outcomes. So that's been something that we talk about a lot, and I think marketing has moved a lot closer in that direction, but I think agents help us get there because we're also automating a lot of the work. Work, that is the stuff behind the scenes that you often have the debates about, and like the BDR and SDR function and how that evolves and how they're capable of doing more and actually upleveling even faster some of them into AE roles or FSR roles. I think there's a lot of opportunity for marketing to lead from the front with the technology that will help produce more effective sellers into the sales organization. So there's a lot of like small use cases where I think starting to bubble up. And I think marketing can again lead that as the tech leader and the thought leader and the thought partner.
Frontline. Yeah, yeah, yeah.
Very much so.
Yeah. It's interesting. I'm like, I agree with you wholeheartedly. I was like, honestly, the lines have blurred for a while now. Like the moment you start getting into the agents, the agents don't ask for a job.
Correct. Yeah. They're all about outcomes.
They're all about outcomes. So it's—
and that's how, you know, we've been trying to move that direction as humans. For a long time is to have shared and aligned outcomes. But the agents start from that point.
They start— yeah, they start from that point. I think something too, we've been, we've been consistent on, on, on, on what we've seen with all the different agents because, you know, we use Gemini amongst other things, is that you have to be super direct with your agent about what its goal is.
Yeah.
And then also giving it the right context. Yeah. Like those two things together, I think, can equal the output that you've kind of seen here throughout the week from some of our speakers is like the number one question, like some of our speakers are getting is like, okay, but how did you really get that to do that with the agents? Because I do, I try and do the same thing and maybe it's not as good. I'm like, well, did you give it the right context? Did you spend enough time on it? Do you have enough, do you have enough clean data? It doesn't have to be perfect, but just some clean data to have a good foundation. I was like, think of it, stop, maybe stop thinking Thinking of it as an agent and just another, not person, but thing that you are hiring into your org that needs context, it needs training the same way a human would on day one. It's just gonna remember the context a lot faster and a lot more consistently than anybody else you bring into the org.
I think that's beautifully said, and it is, but you don't get out of the upfront onboarding and training of it's the same thing as you're investing in an agent. I'm so with you. I think like we in Gemini Enterprise, for us, that's where we're building all of our agents for marketing right now. But the shareability of that is awesome. But like the proliferation of how many of those agents exist has now made us go immediately like, wow, we need governance around, right? Like how do you, how do you start to deploy certain subsets of agents to certain teams so that they know where to begin versus it sort of being, you know, a thousand flowers blooming and all things available to everyone. But I I think to your point, the skillset of providing the context and training the agent and designing those workflows, we can make it as self-serve as possible, but you've gotta be really thoughtful about the outcome you're trying to achieve. And it makes that skillset for marketers at a premium now like never before. It kind of—
100%.
It's like we always also got all this, not all, but a lot of us, you get a lot of gratification in the output, right? And now it's actually The input is the most important thing that you can create so that the agent is responsible for the output. Yep. And I think that's just, it's a shift in how we think about our role and our work, but also our purpose and the value of what we bring every day to the job.
So two last things I wanted to ask you. One was, because you have an insanely driven workforce, do you ever find that you have to rein it in at Google because everyone just has all their agents and they're all using all G Suite tools, like how do you rein it in with folks sometimes?
Yeah, I mean, we originally did not want to limit that. It was for sure, in particular, kind of in the last, like 18 months ago, it was a thousand flowers blooming and that was, we wanted that garden to be really full.
Let it go.
You know, and now we've gone through a kind of a, I would call it a quiet, because we didn't want, we still don't want to stifle innovation, but we've gone through a more behind the scenes structured curation process where we're starting to see what's taking off, what teams are really not just doing a one-hit wonder but actually starting to scale things and doing and applying it to workflows again, not just individual tasks. Yeah, and so, and what also is the most applicable to other teams because we have multiple different types of businesses from, you know, the Chrome business to YouTube to Cloud, all very different businesses, but the marketing discipline discipline still has a lot of similar qualities to it.
Yeah.
And there's a lot of things that if they build it over there, it's gonna be applicable to us. So we've tried to put infrastructure in place behind the scenes that can allow us to have that shareability.
Yeah.
And to then really put the horsepower and the engineering support behind the things we want to scale further for every single marketer at Google and eventually also our customers. So that's been a really important part of the feedback loop for us. And right now Gemini Enterprise sits at the center of that. So a lot of the tools that are being used and built across every single business unit are all built in Vertex, which is now Gemini Enterprise Agent Platform, artist formerly known as. And that is for us really awesome because now everything is built on Google Cloud and we can take that and then scale it across the whole business. So that's been an important part of how we've accelerated it is having the full stack behind the scenes power. Empowering a marketing team.
Yeah. And then my last question would be, any— since Google, I mean, Google Gemini, everything you guys touch right now is on fire. Any parting advice, maybe just in marketing or just in general for the CMOs in the room, of how they could catch a little bit of that fire too?
Gosh, I'm actually envious of people in this room. I think it's such a gift to be, in many of your cases, building AI-native marketing teams, or at least like the next generation. You may have started in a different way, but you're evolving into that. I think that's awesome. Like, I, I think back to when I started at Google 6 years ago, we had so much to clean up. And it was at a time when if I had, if I had a fleet of agents, oh my God, could I have saved time? I spent 3 years in the trenches just like building infrastructure for our marketing organization.
Yeah.
30 different data silos, like bringing them all together. And we're at a really good place now. And fortunately, we have a platform that enables that with our Agentic Data Cloud. But that existing 6 years ago would have saved me probably a year and a half of time. Now, that said, I just think, again, continuing to focus and pick your battles in a good way, pick your workflows as your battle. And I would say leaning into hiring people who are deeply curious, I think you've already got that. You're already doing that. If you're in any kind of startup or early stage company, like that's who you're hiring. So continue to lean into that, but also I don't— I think encouraging the max amount of creativity and that the human still sits at the center of so much of what happens, but to accelerate that and amplify that in any way you can. I think wherever you create that mix of emotion, that's where things are really special and connect with other people and the humans you're still trying to sell to.
Yeah.
So for me, I think it's just continuing to focus and to pick your lane in a good way, pick your workflow and go flex on that.
Yeah. Now let them, let them cook. So the Google Cloud team and I know like the DeepMind and the Gemini team are all in the booth. So if you guys want to see how they're actually deploying some of this internally or how they're doing with customers, go see them in the expo hall. SK, thank you so much for being here.
Thank you so much. So great to be here.
Thank you. Thanks for having me.
Thanks, everybody. I'm just getting a pic of us. Welcome everyone, and thank you for joining us today. Our program will begin shortly. Please make your way to your seats. As a courtesy to our speakers and fellow attendees, please take a moment and silence your devices.
Storytelling. The entire team just focused on one thing: the end-to-end experience needs to be magical. Visual storytelling— there's a lot that happens to make that feel right. Part of building something special and enduring business is doing it with a team. 50 million users a year ago, I would have said, "There's no way. You're crazy." Does this feel like this could be your life's work? And then give yourself enough time to prove that you can accomplish great things. Wow, wow, can you hear me okay? What an honor to be here. Uh, very glad to be here with you all. Excited to share a little bit of our story. We're going to talk about how we achieved $100 million in AR profitably over the past couple years with a team of 50. For those that don't know about Gamma, you can think of us as the anti-PowerPoint. For anyone that spent a late night trying to format, design their slides to try to look nice, we want to help remove all the mundane and tedious tasks so you can focus on your ideas and get them out there. A quick glance at Gamma. We passed $100 million in ARR. Again, a team of 50. and today I'm here to share a little bit about our story. I have 4 lessons, just 4 lessons, hopefully lessons that you can take, bring back to your organizations to help you fuel your growth as well. We're gonna talk a little bit about product-market fit. We're gonna talk about creator marketing. We're gonna talk about community-led growth, and we're gonna talk about dogfooding the future. Now, every early-stage startup has this question of, How do we know if we've reached product market fit? And it's an incredibly important question to answer. Now, unfortunately, this could be one where it's super easy to trick yourself or to fool yourself into thinking you might have it. And so I wanted to share kind of our story to, uh, talk about how we actually approach this ourselves. When we first started Gamma, we were setting out to try to solve the blank page problem. Part of this was informed by my own experience. I was basically at a different startup, we got acquired, I started advising early-stage startups, and one late night when I was preparing for a meeting the next day, I was pulling up Google Slides, trying to get my ideas onto this canvas, and just frustrated by the amount of time it took to sort of format and design rather than on the content itself. and that for me just felt backwards. What if we could reimagine this format from the ground up and really allow people to shape and share their ideas way more seamlessly? So we got to work. I ended up meeting up with two of my former colleagues from Optimizely, and we started building Gamma. Now, our ver— very early version of Gamma was this prototype you're seeing, which is sort of a hybrid document slash deck, and we really wanna make it kind of dead simple for you to sort of flip flip between these different formats. And we spent a lot of time just trying to get this right, where we got to the point where this prototype was our pre-seed pitch deck. We used it instead of using Google Slides. For the first time, rather than using slides, I was able to present with a different medium. In this, in this case, it was our prototype. Now, I was living in London at the time, so this is me trying to practice our pitch, you know, during the daytime. We didn't have a lot of furniture, and so I had to borrow my kids' little desk during the daytime to do all of my practicing. And because we were in London, I had to, you know, do all the pitching in the middle of the night. So come 8:00 PM, I would tuck my kids into bed, I would pick up this little table, move it to the far corner of the flat, which was tucked in between our kitchenette in our laundry room, and I would sit down, put up a fake Zoom background so nobody knew exactly where I was, and I'd start pitching. And, you know, first-time founder, so I'm getting into pitch mode, you know, get through my first two pitches, they do, they go pretty well. I'm like, okay, how hard can this thing be? Get into my third pitch, and 20 minutes in, you know, I finished my pitch and I'm waiting for the investor to start asking me some questions. Long pause, and eventually the investor says, "This has got to be the worst idea I've ever heard. Not only are you going up against massive incumbents, they have ultimate distribution. There is no way you will ever succeed." And so of course I'm a little shell-shocked in the moment. I'm trying to think about, okay, where does— a really smart rebuttal I can give this investor. And before I can do that, he hangs up, leaves the Zoom call. And so in that moment, you know, I tried to internalize what he said because I think there is some truth to it, which is being in this category is incredibly hard. You can't just build a great product, you have to think about distribution from day one. So fortunately, we were able to raise a little bit of money and actually get to work and really think about how do you intertwine design, distribution, and product at the same time. 2 years later, we launched our public beta. For anyone that's building today, that feels like a lifetime. I would advise you to probably compress that by an order of 10 or more and really try to get to market much faster. It took us a couple years. We launched our product on— we ended up winning product of the day, product of the week, product of the month, and we felt really great about ourselves. We're like, hey, Maybe we actually did it. But unfortunately, signups spiked and then they plateaued. And so what we realized is that we did not have strong organic growth. We did not have a product people were telling their friends about. We did not have strong word of mouth. And so we were basically at a fork in the road. What should we do? Should we invest more in marketing? Or should we invest more in product? You know, there's a saying of, "Hey, you should do things that don't scale." And I think a lot of founders, they hear this and they say, "Okay, well maybe that means I can brute force my way into sales and marketing and really just fill the top of the funnel." The alternative maybe is, what if you went back to the drawing board and really thought about the product and really broke the product all the way back down and thought about what would it take to build a product to have much more organic growth. And so the question is, what would you do? For us, we decided to invest in the product. We went back and said, we have to make the first 30 seconds of this product feel magical. It has to be so magical that people tell all their friends about. It cannot just be good enough. And so we got the team together, 12 of us at the time, and really treated this as a sort of all hands on deck. We were crammed in a converted 2-bedroom apartment in San Francisco, and we got to work. We gave ourselves 3 months to re-architect the entire onboarding experience of the product so that we can deliver that magic, that first 30 seconds, so that people would tell all their friends about it. The version of the product was pretty simple in the beginning. You entered a prompt, and you get a first draft of your presentation. From there, you can continue to edit with AI to get you further along on the journey of creating content. Like most startups today, we basically tried to basically launch by announcing this via a tweet. And, um, we sat down, you know, this is again 3 months in, heads down building this product, and we thought about, okay, what would we say if we wanted to announce this to the world? In the very beginning, we had a pretty conservative, safe message, but then we realized as founders, like, okay, sitting down, we needed something a little bit more provocative, right? You need to be able to have something that actually created engagement, and so in many ways, you're trying to lean into this feeling that maybe you're gonna be a little bit clickbaity, but that's okay. You need that initial engagement, so we ended up coming up with this, which is, "The most valuable skill in business is about to become obsolete." And yes, a little bit of clickbait, but it worked. We got Paul Graham to respond throwing a little shade at us, and the tweet ended up going viral. And so it was, for us, a moment of realization, like, "OK, let's get that engagement, let's get that virality." And more importantly, the product went viral. So in the very beginning we had, you know, 5,000 signups a day, then 10,000 signups a day, then 20,000 signups a day, then 50,000 signups a day, and it kept on going up. And this was a moment in time where we're doing zero marketing, zero sales. It was people telling each other, people using the product, that sort of organic growth that we were kind of seeking from the very beginning, and we felt that pull. So really quickly, before I move on to the next topic, And this is the reminder that I'd have for anybody that's building today. Before you spend any money on marketing, build a product that has strong word of mouth. Because word of mouth is the only form of distribution that can help amplify everything else you're doing. Until you have it, I would not worry about anything else. Word of mouth is super hard earned. It cannot be bought. And so go back to building the product that actually has this before you actually worry about the the rest of marketing. We'll talk a little bit now about creator marketing, because for me, that tweet I mentioned, that was my sort of first foray into trying to become a creator. I realized that being a creator, I had a chance to help amplify everything we were doing at Gamma. So I wanted to say, okay, if we know creator marketing can be a channel for us, how do we tap other creators? And I'm a finance person, I had zero idea How to become a creator, how does creator marketing work? So I wanted to learn and I wanted to have deeper empathy by actually going through the journey of becoming a creator myself. So this meant doing the basics, posting regularly, trying to understand the fundamentals of being a creator, learn what it takes. And in doing so, I was able to finally gain a little bit of traction. I tell founders today or anybody in startups, like, if you want to do this, you have to go through cringe valley for a bit. You're gonna put stuff out there that feels just terrible, but that's the whole point. You learn, you put yourself out there, you build the audience. And today I have a little bit of a bigger audience where anytime I announce stuff for Gamma, it gets a little bit more of pickup, and you kind of keep on building that over time. And these are things where I can now take that, what I've learned, into helping us actually build a much more effective Creator marketing program. I can go in and say, when we sit down with a creator, really help them understand how can they tell Gamma's story in their voice to their audience, because I understand the process myself. And I think a lot of people treat creator marketing or influencer marketing as very transactional. They put a budget, they signed up a big creator, and they said, okay, that's it, we're done. We treated it as something where every single creator we would manually onboard I would sit down with them, and guess what? Many of them struggled to create presentations on their own in the past because this wasn't something they learned on the job. And so how could they actually leverage our tool? They ended up falling in love with using Gamma. And so when they go out and tell, you know, their audience how to use Gamma, it was much more authentic. It was something they were actually using. You don't earn the right to do that until you've actually invested in building that, you know, community of, of creators that can actually sing your praises. For us, what we found is that there's just massive benefit. Any time creators would amplify what we were building, there was this halo effect. There was even more word of mouth. So not only were they able to tap their audiences, there was this multiplicative effect of going out, creating content, and it was really creating this flywheel for us. When we think about basically what got us to 50 50 million users, the real big question is for us was like, okay, we got to 50 million users. Do we have everything we need to reach the next 500 million users? And when we're asking that question, we really felt like one key ingredient was sort of missing. We didn't feel like we had a brand that I could actually reach the next 500 million users. And the reason was because our first brand was more of a placeholder. The DNA was very limited, so the number of assets we could create was very limited. And if you want creators or even your own content team to really start amplifying content, you need a much richer set of ingredients, is what I call, like the DNA of a brand. So at 50 million users, it's pretty risky to say, hey, we're gonna do a complete rebrand, because you can cause a lot of confusion in the market. People see, you know, one version of your logo or, you know, your, your color palette, and like you're trying to do something different. We felt like that was the the right long-term bet. Because again, we needed that universe, that DNA that people can then tap into so that they can help amplify our message. And so we did a complete rebrand. We chose a very different art direction, different vibe, something that felt like it really matched our values and what we wanted to be and represent. And we actually even ended up open sourcing this so that at brand.gamma.app, a creator can pull all the assets they need And even more importantly, they can actually create their own gamma-branded assets so that they can further amplify our message out there. And so this was a huge investment, but we felt like, again, at 50 million users, the bigger risk was not to do this. Because if we wanted to reach the next 500 million users, we had to get this fundamentally right. I'll talk a little bit about community-led growth, which is— community, I know, is a term that gets thrown out there all all the time, right? And so what does this actually mean? For us, it meant many different things. One was, you know, we talk about learning from the voice of the customer. We took this pretty literally, which is we actually wanted to hear our users going through the product and struggling along their journey. So we— what we do is use tools like VoicePanel and user testing and actually have users walk through the product and describe everything they're seeing on the screen. And this is where you can really start feeling like, where is their true confusion in using the product, or where is their true delight? And you can start leaning into that much more. We felt like this was fundamental in really understanding how our customers perceived our product. As we started getting more and more power users, we developed a program called the Gambassador Program, which is we have this separate Slack workspace, We pull them in. They get early access to new features. They give us feedback. They're oftentimes the first people to report bugs. And we have this ongoing dialogue. In the very beginning, when we're just setting this up, it's like every single day, I'm the first one to log in to check people responding or asking questions there, and treating that community as one where they can really help us shape the future direction of Gamma. We felt that was super important. The bigger you get— the easier it is to treat your user base as this sort of faceless entity, right? And I think as a startup grows and has traction, you have to do everything you can to resist that. They are not a faceless entity. They are real users with real problems. And so we've always tried to lean much more into this. In September, we ran a program called Gamma Lab, which is we invited users from around the world to work out of our office in San Francisco. We flew them into San Francisco. They could work alongside our team. We could show them what exactly we were working on. And as we were developing our own education program, we could get their immediate feedback so that we can make sure that future users learning gamma could learn it the right way. And so this was, of course, a big investment, a huge time suck because you're like, really trying to carve out time in your day. But again, for us, a no-brainer worthwhile a worthwhile investment. More recently, we had a program called Gamma Everywhere where we visited our users in Seoul, in London, in São Paulo. And again, this was so critical because what you can take for granted is that each of these users, these user bases may need something different out of what your tool offers, right? And so how do you build a product that feels truly native to the users in those regions? You can't do that unless you're not actually spending spending time with them, running workshops, working alongside them. Again, huge investment for us, a no-brainer. The last thing I'll talk about is dogfooding the future. For those that know— are not familiar with the term, dogfooding means you're oftentimes testing early versions of the product yourself before it ever goes into the hands of users. And for us, this has been a critical part of the learning journey from, from day one. I'll share a story, which is when we first started Gamma, our passion, our sort of mission, was to help people change the way they communicate, communicate their ideas, to be more effective at it. And we actually had two different ideas. The first idea was to reimagine presentations as a category, which is very much the product you see today. The second was actually to create a virtual office, and like, what if, you you know, virtual office allowed people to collaborate just as if they were in a real office. And that for us was also interesting, especially in the moment in time where we're coming out of the pandemic, hybrid virtual work was all sort of being, you know, reimagined from the ground up. And so we gave ourselves 6 months running these 2 products in parallel. And when I think about dogfooding, it is all about building conviction for your product. And so we wanted to say, okay, after 6 months, which product did we actually have more conviction on? And more importantly, which product were we more energized by? And it just became so clear with the virtual office, we were sort of almost like there was a ceiling in terms of what we imagined could be possible. We couldn't imagine actually being better than in real life, you know, work. And so we're like, okay, that seems limiting. We can't actually imagine what's needed. Whereas presentations, there was just so there were so many things that we wanted to build, and there was energy pulling us in, into that category. So 6 months in, we ended up sunsetting the virtual office, going all in on presentations as this category, and we continue to do that today. So, Gammon Imagine is a more recent launch where people can design anything, one-pagers, infographics, and we oftentimes are dogfooding these sort of things internally a lot before we actually release it to the world. We've also recently released connectors to different platforms, and for us, dogfooding internally just allows us to see what if Gamma lives in other tools and other workflows where our customers are already using us. What would that look like? How would that feel? So these were just the 4 lessons, hopefully some that either resonate or can inspire something for you all to think about as you bring, you know, all the learnings from this conference back to your organizations. I'll just lead by saying this: for anyone that's building a startup, I'm cheering you on. Honestly, I know there's going to be ups, there's going to be downs, but the, the journey is worth it. It can be daunting. I'm here for you. I would love to be helpful in any way possible. Connect with me on LinkedIn, on X. I'm here to learn as well. So really appreciate the time here. And we're going to do a little bit of Q&A.
Do it.
That was awesome.
Which seat do I get though?
What, what would, what seat would you like?
I'll do this one.
No, do you want the desk? It's like classic SaaS to sit at the desk.
No, that's, that's more of your vibe. Okay.
Okay, I've got a lot of questions. And then for, for, well, you got a big crowd up here. Turned around from the front. I was just saying to you backstage how I used Gamma last night to create the little Pokémon cards I made for our opener this morning. And there's not only are they so cute, it did it instantly. And when I gave it feedback, I didn't even say in the prompt, make me Pokémon cards. I was like, hey, I want to make our AI agent stack and I want to just visualize it for everybody at Saastr because this whole week Amelia, Jason, what's your stack? What's your stack? Show us the whole stack. And I was like, okay, I'll show you. I'll try and show you some of the stack. I don't even know if I can show you all the things because we have like 30 agents. And so I just went into Gamma and I said, hey, here, here's the list of my agents I want to show. Make me something to show people the stack. And it was Gamma Create and it asked me like, that was literally the prompt. Just make me an image to show people the stack at SaaStr. And it said, do you want it like Western style, like wanted posters? Do you want— there was some like more like infographic style in the middle. And then I was like, or do you want Pokémon cards?
Like, no brainer. Pokémon.
No brainer. I was like, got to catch them all. So that was super fun. But I just— I've been using Gamma for a little bit now, both myself, but also one of my agents also uses Gamma. So one of the things I want to ask you, because we talked about on the prep call, was like, you've built such a product that is, you know, for humans. Being creative. But I know you were saying though, like, you have now more customers that are using the Gamma API. So like, how are you balancing that, right? When we're all here at SaaS, we're living a world now where like, I use Gamma, but also my agents use Gamma. How does that impact what you're doing now and in go-to-market?
I mean, I think this applies to almost most categories, which is everyone building today needs to have kind of those two mindsets, right? How do you build for the human, as well as prepare for a world where many more of your users are agents. And I think that's a journey we're on today, which is we do want to serve both. We think they can be deeply intertwined because a lot of our biggest customers do have both end users using Gamma as well as a ton of automation leveraging our API. And so it's part of the fun. It is sort of changing the mindset though. And you almost have to go from like the fundamentals, like, How do you make a product that could hopefully be great at both, but also maybe not alienate one or the other as you're trying to innovate along the way?
Yeah, and then another thing I wanted to ask you, and then we'll do some of the questions I came up with earlier, but when you were talking about your gambassadors, right? So you have users around the world that you've brought in, you put them together. How much of that are they asking you for features they want and then features their agents want?
Today, it's still very heavily features they want. Because I think, you know, the reality is most people are still— I think this has brought like one realization coming back from this world tour was that, you know, we're still so early days, right? Like obviously being in San Francisco, being in the Bay Area, we have front row seats to the latest and greatest. And of course there's almost a little bit of an echo chamber here. But when you look broadly, one, the market's super fragmented. There is no, you know, not one person, one, you know, team or one one, product is dominating everywhere, and two, the sort of, the adoption curve, like most people are still stuck at just getting the basics right and trying to understand like how do I actually fit this into my work life. And so today a tool like Gamma slots in and I think most people are just trying to even take their first step into being more of an AI forward or AI first thinker. And many of the features they're imagining are ones where we can just make it a little bit simpler, that first step a lot simpler.
Yeah, so you guys got to the first $100 million, no sales team, still no sales team, or some sales team now?
Growing sales team, actually some of our folks here, yeah. Growing sales team, you know, our first part of our journey was really just this prosumer motion, which is still our core growth engine, but we recognize like our ambition is to build a product that could be ubiquitous, right? It could be a new standard, and to be a new standard, you not only need to win individuals or small teams, you need to be able to fit in into a much larger context of an organization. How do you win businesses at any scale? And so our sales team is here to help make sure that we can fulfill the second part of this journey, which is actually being a core business tool across organizations of all sizes.
And are you still just— some of the debate here this week has been, do you still distinctly hire a sales team, a marketing team, or since everything's becoming blended, do you just hire now for go-to-market where they do a little bit of everything?
We're still at a stage where, you know, we hire different teams. They're probably much leaner than they would've been, you know, 5, 10 years ago, and they're highly collaborative, right? There is that tight feedback loop between what is a customer hearing, or what are you hearing from customers to how can we ensure we're talking about the product the right way? And of course, product is another team that has a seat at the table to ensure, hey, we are trying to build the future for them. Let's make sure we're building in a direction that solves real customer problems. Customer problems. So it just ends up being super flat, very collaborative. Teams are probably much leaner than they used to be. And of course, you're trying to plug into your own context engine to have all the data, all the customer transcripts, and making sure everyone is informed about where things are going.
Yeah, and since you started with a PLG motion, where did you inject the sales team? So is it just enterprise? Who gets to I wanna talk to a human now.
Yeah, this is definitely stuff we're still evolving. In the very beginning, you know, we have the benefit of having a ton of inbound already. And so just fulfilling inbound, we haven't even really begun the process of, you know, all of our self-serve customers, like actually going and engaging them. Of course, that was probably gonna be sort of next phase. And then eventually, you know, you're gonna be doing much more outbounding. We are candidly like early on the journey and like really trying to think be thoughtful about, okay, we have, we believe this massive opportunity in front of us, how do we sequence it the right way?
So I know it sounds crazy, but for folks that are dealing maybe with similar issues or they start to ramp and they also get too much inbound, like how do you have all humans on that? Do you have any agents on inbound?
We have much more, so it's a split right now. We have both humans, but Then, you know, agents, AI can help us also prioritize the right conversations and ensure that they're getting some touch. Right now, oftentimes not— that's not even the case. So I think, you know, we're working through like, how can we give, uh, ones that maybe aren't as qualified some level of touch, ensure that we can proactively go out to the ones where they, they definitely, you know, could deserve more support. So it is just a chance to like— obviously the amount of information you have today, you can actually do much more of than you could we would've in the past, and we're in the process of trying to really automate as much of that as possible.
And how does that bleed into onboarding, right? Because now you have a little bit of a sales team. How much of it is they can learn Gamma themselves and proliferate to the rest of the team, and how much are you guys trying to get them to use it more?
So one thing I've also observed, and again, this is kind of when you just think macro, like what's going on and how early we're on the journey, I think a lot of companies, customers, business buyers really appreciate a very consultative approach to how do you get the most out of AI. And so for us, these are things like, we have a brand, and how do we translate that into a brand within Gamma? But not only that, how do we elevate it beyond just what the brand says today? And so there's a lot they're looking to us to help them, again, usher in a new era for organization, and many of them are very much leaning into, like, we want the right partners, not just the software, but true partners in all this. I think that's just something we're recognizing is like that almost becomes like an essential ingredient today.
And do any of your customers get an FDE at Gamma?
Oh yeah. Well, so we have different versions of basically either like a, what we call a forward deploy designer, someone that can actually help them with some of the design translation translations, as well as, yeah, more technical folks that can help set up different things like our API, for instance.
That's pretty cool.
You guys also hit 600,000— correct me if I'm wrong— 600,000 paying subscribers. That's a lot at 100 million. When did you— we've talked about it a little bit, but when did you realize, like, at what customer were you like, okay, we can't just be self-serve anymore, we need to have some humans on the sales side?
I mean, it really came to when the inbound was just, you know, we just felt like we were dropping so much. There's so many people asking us for, how do I get my entire team or my entire department?
Yeah.
And we, we felt like we were, we were just missing a huge opportunity. It, for us, in many ways, you know, we've, we've always, uh, for better or worse, been sort of reacting to the sort of inbound or like the situation we're at, and I would advise maybe not do that. Like, you should probably, you know, try to be more proactive if possible. But even with us, when we first launched our self-serve product, we launched— I mentioned kind of like the, uh, all hands on deck, just getting something out there. We launched and we didn't even have monetization figured out. There's no way to pay for the product. Oh. And so we, we had this credit-based system, and all of a sudden, you know, chat was blowing up saying, how do I buy more more credits, and we had no way of actually charging for credits. And so we spent a couple of weeks of actually figuring out pricing and packaging, again, reacting to the situation rather than trying to be a little bit more proactive. And similar stuff happened on the B2B side.
So actually, I want to talk about pricing a little bit, actually, if you don't mind. Because what I noticed this morning when I was making all the Pokémon cards, it was like, generate for, I don't know, like a lower amount of credits and then a higher amount. I think I did the higher amount because they looked good. But like, how do you think about that when the base price is— I mean, for all intents and purposes, pretty low to be a Gamut user.
Yeah.
Pricing and packaging is another area where I think startups in the past used to say, hey, we'll put something up, set it and forget it. I don't think you can do that anymore. I think pricing and packaging is something you need to revisit constantly. And so, you know, today I think many startups are going on a journey of how much should be seat-based versus consumption-based. Consumption-based, usage-based, versus an API that is all completely usage-based. Candidly, we're still going on that journey and we're doing our best to align price to value, which is the fundamentals of pricing and packaging, and also doing it in a way that is sustainable. I think some startups go down the path of, "We will sell at all costs, negative margins." We've run a profitable business, so we want to make to make sure the exchange of value also ensures we can do so in a way that is thoughtful to the underlying business that we have.
And what does it mean for you guys on the backend, right? If me as the user, I'm like, okay, I'm gonna pick the higher model. Like what is that doing on the backside for you guys?
Yeah, a lot of the, in the background is basically the orchestration. So we do work with a lot of different model providers. Each model provider does have, you know, different tiers of models as well. Yep. And so if someone is going for something something that is high polish or maybe is a high-stakes presentation, they really want it powered by the very best, we'll direct them to that path. It's going to be more expensive, but the output's going to be—
Better.
For a higher-stakes presentation, way better.
And how do you tackle speed? Because I was just talking to Google about this. I feel like when I've, in the past, when I've picked higher or lower on that, it doesn't seem to affect, though, how fast it's outputting.
Yeah, a lot of stuff underneath the hood to really try to optimize all of that. So depending on what we can sort of anticipate someone needing, we can obviously hopefully direct people down a direction that feels very fast to the end user, but of course doing things in the background to make sure we can actually deliver on that.
Yeah, just to dive in a little more, I think it's interesting for folks, like when you use something like a Nanobanan or like a Higgs field, sometimes there's like a lag, right? Of like, OK, if you use higher credits, you can sometimes skip the line. Like, for you guys, I'm like, how is it always instantaneous? Do you just have like a server for it? Like, what are you doing?
I don't know if we're doing anything super crazy there other than I think some platforms that have to obviously video completely different.
Takes a long time, yeah.
And depending on how much they're relying on trying to intelligently switch between different model providers, maybe there's some logic there. And in some cases, I think certain products try to actually slow things down so they can charge more for the fast version. We don't do any of that stuff.
Okay, so you just unleash the models directly?
Just unleash, yeah.
Wow.
What is your, like, you guys must have high usage costs for the models that you're using.
High usage costs, but again, like running profitably. We've always tried to run in a way that is strong gross margins because we believe in building a sustainable business.
Yeah.
There's always the balancing. Like, you will do a disservice to your customers if you go out of business because you're running unsustainably.
For sure.
And so you have to make sure you can build up not only a great product, but a sustainable business.
So you're an AI native—
for all intents and purposes, AI native company. Now you have a small but growing go-to-market team. How much of the AI is on— like, what does your go-to-market team use now for, like, their AI? Obviously Gamma, but like what are the other connectors and agents that they're building internally?
I mean, they're constantly experimenting. I mean, like many, like Cloud Cowork, you know, has been a place where obviously there's just becoming like a place where a lot of the context can live, connecting to different, all the different tools, everything we're using, communication tools, Slack, Notion, like everything's connected. And then of course Gamma, you know, we pride ourselves in being also our top, users, talk dogfooters, and using Gamma constantly. So these are all tools that our go-to-market team is leveraging every single day.
When it comes to hiring at Gamma, what is that process like on the go-to-market side versus like the engineering side?
It's, you know, we hire a lot of strong generalists. I think generalists that can spike, of course, in different areas. But, you know, the whole mindset, and I think this applies to engineering too, is like the whole mindset of There is no job too small. We're all sort of— the benefit of being a very lean team is that each person has just so much ownership and doesn't need to be told exactly what to do. There's just like a problem space of things that can be done, and we arm them with as many tools as they need, and we trust that they can, you know, make a lot of progress. And of course, because it's so collaborative, they're oftentimes sitting alongside product team or engineers to actually make meaningful progress on, on any of their our core initiatives as well.
And now that you're— so you have, I think you just said you have one office here that you're— that's growing, and then now you're expanding internationally.
Yep, planning to open up a London office later this summer and then a New York office very, very soon.
And then one thing I said to you backstage was like, every— but everybody on— everybody on your team at Gamma has like this really great vibe. Like, how are you going to tackle that now that you're going remote?
That's a great question. I mean, one of my pieces pieces of advice for any fast-growing startup is in the early days, you know, we had a saying which was hire painfully slow. And part of that is really just trying to ensure you're locking in the DNA of the company. You know the values. You know everybody there shares the same principles. And once you do, once you have that, you start building some critical mass of people that are all like-minded, share the same ambition, same vision, then you can scale up much quicker, right? Because again, you've locked in on the DNA. The people hiring the next set of people aren't going to let someone in that doesn't match the vibe, doesn't match the, you know, the, the same principles. And so that's where I have much more confidence. Certainly part of, you know, expanding, uh, internationally, we're sending Diana and some other teammates to London because we want to make sure our cultural carriers are there. And as they're, you know, building the team there, they are finding people that emulate the same vibes as well.
Yeah. And have you guys decided yet how much of that you'll let Diana do? So she's going in person to do that. Do they still have to meet with you before they get the job, or? I don't know yet.
For certain roles, maybe. But I also intend to spend a lot of time out there with her. But of course, most of the roles will probably be her leading and actually making the decisions there.
Yeah, it's crazy. I mean, go-to-market has changed so much just in marketing, just between sales, marketing, and all the blending, just even in the last year. Last year when like we did SaaStr, I was like, we weren't even talking about folks like creating in the way people can create on Gamma, right? Like, it's totally— it's been a full whatever, 180. But even now, like, I'm like, I use it a bit, but like, I think my agents use Gamma more than I do. Like, they use the API a lot more than I do. Um, I'm like, sometimes I forget to log into it. Like, I basically have like all these headless apps now, um, running around. And so one of the things I I wanted to ask you is, as you have more users like that, how do you instill some of what Gamma has built around brand and almost culture with them when it's an agent, right?
Yeah. I think every company today has to just be super adaptable. I think that is something we also hire for is how willing are you to both learn but potentially unlearn everything that's built your career to this point. So it is hard. I think part of being able to dogfood the product is actually just having that intuition too, because we have so many power users of Gamma internally that are manually creating slides and presentations. That may not be the same mix going forward. We might have a ton more automation. We won't build good intuition unless we're doing it ourselves. So going back to why dogfooding is just so important, you need to have that deep level of empathy to see where are things where are things even heading? Are things even valuable to your own internal team? And if you can do that for your team, I think that's a good, hopefully, signal for what others are going to need as well.
The last thing I want to wrap on for folks is like you guys are in a space that you've kind of like carved out your own piece of a very crowded market, right? Like there's a lot of other folks who do either portions of gamma or like some pieces of— I think some people still have to use PowerPoint because they're forced to use PowerPoint. Like, how have you guys— but you've grown so fast and with such virality, to your point. Like, any parting advice for folks that want to try and— well, I know you said a lot of it already, but like, any last words of advice if they want to try and duplicate kind of the recipe and keep it consistent?
Yeah, I think there's a few things that, you know, just general advice, because, you know, everybody's journey is different. Advice number one is just make sure you really think about about why you're building the company you're building. Because even for us, you know, we're 5 years into the journey and you have to sort of stay committed to like, where are you even going, right? It's so tempting today, given how much AI is changing, to kind of chase shiny objects. You can go and say, hey, I'm gonna bolt this on, I'm gonna bolt this on, keep doing more and more. But when you think about why are your customers choosing you, why are you building your company, what is your long-term aspiration? Aspiration for, for, you know, what you're delivering back to your customers, you need to stay grounded there. And hopefully that also just becomes a massive flywheel for you because no one else is going to be willing to go as deep into that problem as you. Everybody else might be chasing, you know, becoming a super app or trying to do all these different things. If you can stay grounded on what's important to you, I think that just gives you longevity, longer chance. Uh, and then the second question, like every, you know, product builder should be thinking about is like, how does your product actually grow? Is it by use of the product, or is it some other dynamic that as you grow, as the product gets used, is there some sort of flywheel that happens? And like, double down on that flywheel. Yeah, make sure that every step of the way, like, if you're building a product, like, you're removing friction from that flywheel spinning faster and faster. And that's just, you know, sometimes people just get caught up into trying to compete with a competitor, or like trying to, again, like bolt on the latest and greatest when they forget, like, what does that product loop actually look for your own product?
Last follow-up question, um, how do you decide now what gets into the product at Gamma?
You know, I think people used to have like 1 or 2 year roadmaps, right? No way.
Now it's like 2 weeks.
Yeah, so it is, you know, part intuition, part customer input, and then part just seeing like What's going on on a macro level? You know, what are the things when it comes to like AI where innovation is happening so fast that you cannot afford to ignore? And I think those three ingredients are always at tension, but it's your chance to kind of be the cook and go to work.
Well, I know you guys want to ask Grant some of your own questions too, and I think he's going to be in the beer garden afterwards that they're in. So if you guys want to ask him questions, you can find him there. Obviously follow him on social. But thank you so much for doing this and being here.
Thank you.
