The 4 Levels of AI Marketing (Where You Rank & How to Level Up)

0:00:08 I got the train to London today to meet with one of the top AI people in the world who you
0:00:14 need to hear from. Kristen is the head of marketing at Gamma who just hit a $2.1 billion
0:00:21 valuation last week. She’s got this amazing framework for how you can succeed with AI.
0:00:27 While everyone’s obsessed with prompts and chatbots, she’s using AI to think to diagnose
0:00:34 million-dollar marketing problems in 30 minutes flat. She’s trained AI to write like humans,
0:00:40 not robots. She can turn a messy brainstorm into a polished presentation in five minutes.
0:00:45 And here’s the part that actually broke my brain. When I tried to schedule this interview,
0:00:52 her assistant emailed me back. Turns out it wasn’t a person. It was an AI agent Kristen built,
0:01:00 one that books 40 meetings a week. Kristen isn’t just using AI. She’s architecting an entire AI-powered
0:01:07 marketing system. It’s what helped Gamma become one of the fastest startups to hit $100 million in
0:01:14 revenue profitably. And today, she’s breaking it all down. Her four levels of AI mastery,
0:01:21 the cloud project that replaced a data analyst, and the one mistake with GPT-5 she’ll never
0:01:27 repeat. This is the next wave. I’m Maria, coming to you from London, where the future of marketing is
0:01:33 being written by humans and AI working together. My conversation with Kristen starts now.
0:01:40 Kristen, welcome to the next wave. Thanks for coming today. Thank you for having me. Of course. For people
0:01:46 that don’t know who Kristen is, Kristen is the head of marketing at Gamma, and she’s built and scaled
0:01:53 programs that actually convert, not just look amazing on deck, by the way. She’s one of the rare marketers
0:01:59 who can speak brand, product, pipeline, and experiment design fluently. Thanks for coming today.
0:02:04 Thank you so much. So happy you’re here. I’m just going to start off with the first question. You use
0:02:10 AI every single day, like all of us. Of course. What’s one workflow that you genuinely panic without?
0:02:16 Yeah, so one of the workflows that I have come to rely on, I would consider myself to be a slightly
0:02:20 technical person, but not a very technical person. And so data analysis has always been one of the
0:02:26 things that’s a weaker area for me. But I have found that Claude to be an amazing data analysis partner,
0:02:33 for sure. So anytime that I am trying to dig through trends and data, I will also pop it into Claude.
0:02:38 And of course, you don’t want to take anything as truth, but it gives me a lot of good hypotheses to
0:02:43 dig into. So for example, a little while back, we saw a drop in our organic traffic. And I used to work
0:02:49 in SEO and specialize in that field. And it would take me a long time to slice and dice data by country,
0:02:54 by keyword, to really diagnose a drop like that. But I just took our Google Search Console data,
0:03:00 our Google Analytics data, popped it into Claude, asked it to identify the hypotheses for why it
0:03:03 might have happened. And that gave me five leads to dig into. And so within 30 minutes,
0:03:07 I had an answer as to what happened, whereas it may have taken me hours, even days before.
0:03:12 Oh, yeah, yeah, definitely. Honestly, Claude is doing amazing these days. So you said you treat
0:03:18 Claude plus ChatGPT like a thought partners. What does it look like in your day-to-day life?
0:03:23 Yeah, so I would say when I’m at work, I have Claude, I have ChatGPT open at all times. I think
0:03:28 one thing when you are using AI tools a lot, you kind of develop preferences, right? You see their
0:03:31 quirks, you know which ones are better at which tasks, which is why it’s so important to tinker
0:03:34 with them all the time, because the models are always changing. Personally, you know, in ChatGPT,
0:03:39 we have GPT-5. Now, I always toggle it back to 4.0 at the moment, because I have found personally
0:03:41 that creates better writing outputs for me as a marketer.
0:03:46 It’s nicer. It’s nicer to us. Yeah. It’s not robotic a lot. I know. If I forget to toggle
0:03:49 it off, I know, right? Models have different personalities. So they’re always there with
0:03:56 me. I’m constantly asking questions. One of the things that I do all the time is I use ChatGPT
0:04:01 and Claude projects and Claude skills. So Claude skills are probably my new thing. So I have multiple
0:04:06 Claude projects with Claude skills trained on all the things that I do all day long. Think about
0:04:12 brainstorming for campaigns, think about positioning, think about growth experiments. And one of my
0:04:18 favorite Claude skills is one that I trained on anti-AI writing habits, which was actually a really
0:04:24 interesting anthropological study into what the AI thought AI writing habits were. It didn’t come up
0:04:29 with M dashes or rocket ship emojis at first until I prompted it and said, you’re right, you know, for
0:04:35 sure. But lots of things over parallel structure, you know, cliche hooks, like business jargon. And so
0:04:42 it’s not perfect. But I think all of us who use AI now are pretty good at identifying AI written content,
0:04:47 and I would say it takes it up a notch. So anytime I’m doing a writing project in Claude, it’s always
0:04:53 pulling from my human copywriter Claude skill so that it sounds a little bit less AI.
0:04:57 Before we continue, though, a very exciting thing happened in Gamma today. I forgot to ask you about
0:05:04 that. You announced Sirius B, which raised a 2.1 billion valuation. Can you tell us more about this?
0:05:10 Yes. So today Gamma announced our Series B. We have a $2.1 billion valuation. Gamma historically
0:05:15 has not raised a lot of funding. So this was a big moment for us. We are a profitable company. We’re
0:05:19 probably one of the few profitable AI startups. We’re probably one of the fastest companies ever to get
0:05:25 to 100 million ARR profitably. And so it really has been a rocket ship. So super proud to share this
0:05:31 with the team. That’s awesome. That sounds so good, honestly. You talk about levels of AI for marketers.
0:05:36 How can someone tell what level they’re at and what’s an easy step to level up?
0:05:41 Yeah. So it can be really confusing. Everybody’s throwing AI advice at you all the time. So a simple way that I think
0:05:48 about it, there’s four levels if you’re a non-technical person. Level one is prompting. So using ChatGPT and Claude
0:05:53 tools like that to answer questions, get work done faster. Level two is what I call prompting with context.
0:06:01 So that is using things like Claude projects or ChatGPT projects, Claude skills, where you are inserting some
0:06:09 underlying system prompting with data that you have. So for example, we have a project that is trained on
0:06:14 our brand voice and tone. And so that involves putting in our brand guidelines, our style guidelines,
0:06:20 our marketing playbook, things like that. So anytime that we are prompting in order to figure out something
0:06:24 that should go within a particular marketing initiative we’re doing, then we use that project,
0:06:31 right? And that will uplevel your results. Level three would be vibe coding and prototyping. And so
0:06:36 one thing that our creative director does on our team is she never comes to us with just an idea. It’s
0:06:41 here’s what it would look like. I vibe coded this game. So this summer, we released this game called
0:06:46 Escape the Office, a super fun, and this runner game. And in order to pitch it to the team and to get
0:06:53 sign off on it, she prototyped it in a lovable, I think. So vibe coding, prototyping. Level four would
0:06:59 be agents and automation. So creating your own agent to get stuff done, tying together different tools
0:07:05 and automation. So basically letting the AI work for you. There’s level five and level six for the more
0:07:08 advanced. And I’m sure as AI keeps developing, they’ll be there. But I think, you know, if you’re
0:07:13 just a general non-technical marketer who wants to be AI first, trying to get to level three or level
0:07:17 four is where you want to be. Everybody should be beyond level one right now. So at least use cloud
0:07:26 projects or chat GPT project. That’ll guarantee you get much better. Yeah. Yeah. Yeah. All right. So if
0:07:32 you’re wondering why your marketing playbook suddenly stopped working, it’s not you. AI has literally
0:07:38 rewired the game. The old rules are just dead. That’s why we’re giving you the AI marketing rebuild
0:07:45 toolkit. It includes 10 prompts to fix your traffic, ads, email, and video strategy for how people
0:07:51 actually search and buy now. If you want to level up and adapt before everyone else does scan the QR code
0:07:55 or click the link in the description. Now let’s get back to the show.
0:08:03 To follow up on that, when it comes to using cloud, you use cloud as a growth analysis partner,
0:08:09 which that’s badass. What’s one insight it found like you found that surprised you when it comes to
0:08:15 this function? Yeah. So we use cloud all the time. I put tons of data into cloud. So there’s lots of
0:08:20 tools that you probably have on marketer that have robust amounts of very granular data. Google Analytics
0:08:26 is one of them, Google Search Console. And so if we’re trying to diagnose an issue with organic traffic
0:08:32 or things spiking in certain countries, we can plug that data into cloud. And what I try not to do is be
0:08:37 leading with it. I just say take a look at this data and tell me what you see or something along those
0:08:41 lines. And then kind of get the trends and then start digging in deeper into hypotheses. And so if we’re
0:08:46 trying to figure out something dropped, then we can do that. It’s also great for identifying opportunities
0:08:50 that you weren’t aware of. And so on a weekly basis, I’m usually plugging some data into cloud just to see
0:08:54 what it comes up with. And then, oh, maybe I didn’t realize that opportunity. Oh, something spiked in
0:08:58 Japan. What’s going on there? Oh, maybe I should look into, you know, some influencer campaigns or like TikTok
0:09:03 videos that took off. And so even if you’re not trying to dig into a problem, just if you
0:09:08 have robust sources of data that you can plug in and just ask Claude or, you know, ChatGPT I’m sure
0:09:12 works as well too. Just what do you see here, right? It can give you a lot of ideas.
0:09:18 That sounds so cool. So let’s go back to Gamma right now. Gamma built a prompt guide by studying
0:09:24 successful user flows, which is awesome. I think what people want when it comes to their workflow.
0:09:27 What’s one underrated tip most people don’t know about?
0:09:32 Yeah, so Gamma’s AI product has been around for two years. And one of the most common requests we’ve
0:09:37 gotten over that time is give us a prompt guide, right? Somebody said, “No, Gamma’s supposed to be easy.
0:09:42 You just, you know, put anything in and it’ll help you,” which is true to an extent. But because we have so much
0:09:48 data and so many active users, we have looked into what the top 800,000 active users are doing with
0:09:53 Gamma. And we separated them into different, you know, job function personas, marketers, consultants,
0:09:56 to see whether they’re successful. So it looks different, right? It looks different for each
0:10:00 of the prompts. You can look at our site and kind of see how a consultant prompt looks different than
0:10:04 a teacher prompt, for example. But one tip that I would give to anybody that’s completely underrated
0:10:11 is asking Claude or ChatGPT to write your prompts for you. And so typically, I always start by giving a task,
0:10:16 right? I want to ask you to do this. Here’s my task. What’s the perfect prompt for you? And I want
0:10:20 this type of output. I want it in a chart. I want it in a table. I want some graphs. And then say,
0:10:24 how would you prompt yourself, basically? And then you’re going to get a much more robust prompt,
0:10:27 and it’s good enough. It works for now. This is actually something I have got to give a hat
0:10:34 tip to Kieran at HubSpot for teaching me this. Shout out to Kieran. So if someone’s new to Gamma,
0:10:37 what’s the easiest five minute win they could try today?
0:10:46 I mean, take a document that you have that is a bunch of text, or maybe even the transcript from a
0:10:52 Zoom call, and just put it in Gamma and get a beautiful deck. One of the reasons why decks have
0:10:58 been the language of business for so long is because we are visual creatures. We’re multimodal. We can process
0:11:05 things a lot faster. And so you can take blocks of text, pop them in Gamma, immediately get something
0:11:11 that looks way more presentable and more appealing to share. I like to automate this with our API. So
0:11:17 my AI note taker app of choice is Granola. We also use Fireflies. We also do Zoom calls, transcripts.
0:11:20 You can connect that to Gamma. And so anytime you have one of those tasks where it’s like,
0:11:24 okay, after the meeting, here’s our recap and next steps, right? We can do that for you.
0:11:31 So with the API and Gamma, note taker app, prompt it, give me my next steps. In Gamma,
0:11:35 you’re going to get an automatic deck after the meeting. So you can share the recap with anybody
0:11:39 who missed it or next steps. Or if you’re talking about an external call, you can immediately impress
0:11:44 them with a follow-up deck that perfectly curates everything you talked about within five minutes.
0:11:50 That sounds so cool. To finish it off, I want to talk more about women in tech. Let’s give a shout
0:11:55 out to the women in this workspace. I really want to talk about that, especially in Gamma. Can we do
0:12:00 that? Yeah, absolutely. But before I do that, a little story. Many years ago, when I went to college,
0:12:06 I studied aerospace engineering. And so my freshman year… Can I get this shock on my face? I’ve had a
0:12:12 storied career. Oh my God. Yeah, I’ve done many things. I switched to being a theater major. So, you know,
0:12:16 that’s fine. Still, for me, that’s badass. Yeah. So I started saying, I interned at NASA,
0:12:21 I wanted to be an astronaut. So I started studying aerospace engineering. 18 years old freshman. I went
0:12:27 to a big state school. We’re all sitting in the orientation. There’s probably 300 kids or something,
0:12:32 probably like a handful of women. And this professor gets up in front of this entire group, a very
0:12:38 impressionable college freshman, and said, “Most of you girls are going to drop out.” And that stuck with
0:12:43 me forever. And it stuck with me at that time. I said, “I am not going to drop out. Like, no way. I’m
0:12:47 getting through this.” Well, it turns out I didn’t like aerospace engineering as much as I thought I
0:12:52 was going to. That’s okay. We always change stuff. I did change, but look at me now. You know,
0:12:55 working at Gamma, working in AI, staying at the forefront. You’re doing pretty good right now.
0:12:59 Staying at the forefront. But, you know, it’s really important for women to lift each other up. And so
0:13:04 one thing that I take inspiration from, I wish I had the exact counts before this interview, but I
0:13:09 believe Gamma is about half women. So the head of our AI product is a woman. The head of core product is a
0:13:14 awesome woman. We have women AI engineers. We have our own, you know, a group, our own Slack channel
0:13:19 to support each other. And so even though it may look a little skewed in the general industry,
0:13:24 really surrounding yourself with a community of supportive women that build your confidence is
0:13:29 super important. That sounds so cool. And I’m so happy to hear about your journey. That was insane.
0:13:33 Thank you so much for coming to the next wave. And hopefully we’ll get to meet you soon again.
0:13:38 Yeah. Thank you for having me. Of course. If people want to find out more about you and like
0:13:42 follow your work, where do they reach you? You can connect with me on LinkedIn. That’s my social
0:13:47 platform of choice. Okay. That’s perfect. Do you have any sort of work that people like a portfolio
0:13:52 people could look at? I mean, my life is gamma.app right now. So, you know, check out Gamma,
0:13:57 please. Cool, cool. Thanks for tuning into the next wave. We’ll see you next time. And don’t forget to
0:14:10 like and subscribe. Bye.