Generative AI & the CMO
Yoav Rimon: My Growth Manager in my team, he told ChatGPT how to build a table based on different criteria in different columns. He generated about 5 pages in less than 30 seconds.
In one day, the market has transitioned from being a nice, cute niche to the leading example of AI in the world.
Becky Holland: Hello and welcome to the No Bull Marketing Ideacast. I'm Becky Holland.
My guest today is Yoav Rimon. The VP of marketing at Brew. They’re a marketing tech company and we’ll find out a little bit more about them as the conversation goes on. But he was previously CEO and Co-founder of his own mar tech start-up, and an ex-Googler, if you like.
So, what we want to try and do today is just talk a little bit through Yoav’s story. Find out about where he came from, what it was like being the CEO of a mar tech start-up and really where he's coming to today.
We're also going to talk a little bit about generative AI because that's clearly the topic of the day and something I know I'm pretty much not without day-to-day even just a couple of weeks in. So, without further ado…Yoav, good morning, welcome to the show.
YR: Hi, Good morning. Thanks for having me!
BH: You started out at Google, right? How did that come about and how did you go from there to starting up your own mar tech business?
YR: Yeah, good question I mean Google wasn’t my first job, I did have a couple of jobs before that doing marketing at start-ups.
YR: I didn't know I was going to do marketing and they don't really know what I was going to do at all. But then I had this opportunity to sort of interview for a start-up for a very general description of someone who “will do marketing in some way.” I remember, during the interview the person who was interviewing me, the manager, he showed me Google Analytics and Google AdWords.
YR: …and that's when I knew “OK this is where I want to be”. I was amazed. I didn't know that you can actually see the traffic that goes into your website. I didn't know that you can actually create ads that will appear in search results.
BH: Do you know I think a lot of people still don't know. Still don't know that today. “How did those ads appear? How do they know?”
YR: It was roughly 16 years ago. Or 17? I don’t remember. So, then I had this…I had a few gigs and I saw a job at Google. Went through the process. I relocated to Dublin - from Tel Aviv to Dublin - with my wife. We stayed for a couple of years and then I went back to Israel and continue working at Google in Tel Aviv for another four years.
BH: So, what were you doing for Google?
YR: Yeah so, I was working with the biggest advertisers in Israel and a couple of other regions helping them build their strategies with the Google marketing tools or Google AdWords or Google Ads as they call it today. My focus was almost always on the most innovative features of product.
So new display products, mobile, and other more- mostly performance oriented - features and tools. So, I would educate customers, I would build strategies for them, work very closely with CMOs and marketing managers from the Google standing point, it was, you know, bring in more budgets but then the customers would get a lot of value from working closely with the Google. They had to be spending quite a lot to get that sort of service level from Google as you can imagine. Yeah, so it was fantastic six years of my life.
BH: It must have been strange for you to move from Tel Aviv to Dublin. Did it feel like a big shift?
YR: It was it was I mean it was like my wife and I were engaged. I didn't even tell her that I submitted my CV to Google…
BH: Oh wow!
YR: Because I didn't think they would reply, and then we were driving in my car to her parents’ house, and I got the call, and it was back when we…people didn't have the vocal system that anyone in the car can hear so I had to listen [in this] like little headset.
BH: Oh wow!
YR: She's sitting next to me, and she listens to me speaking in English. She didn't know who it was, and then I finished the call and I said “they called back from Google. I submitted my CV.” She's like “OK that's great” and I said, “yeah but the job is in Dublin” and she started laughing.
BH: I bet she did, and she still married you!
YR: She said, “that’s hilarious I mean we're not moving to Dublin” and then we moved to Dublin.
BH: Eventually you decided that it's time for you to go and do something else. So, tell me what happened. What did you do next?
YR: After a few years I felt like “okay I'm 30 something” and I said, “I'm not going to have a lot of opportunities in life to go out and just do something on my own” and then I was connected with another guy who was also searching for the same thing. But he was technology person. So, we hooked up we had a few meetings and then we decided to just go out and launch a start-up. Funny enough it was in the days, a few years ago, when chat bots were all the buzz.
YR: Everyone was talking about chat bots around between 2016 or so. But AI wasn't yet ready. I remember pitching to VCs and investors telling them “Look all those start-ups that are trying to raise money saying that they have AI, that knows how to have a chatbot really talking to a person, that doesn't exist yet. So, we're going to do something similar. We will use some of these capabilities, but the market isn’t ready yet, but we do have a solution for marketers to build chatbots for the businesses and then having them emulate a conversation and then just try to convert leads better or retain customers better using automated chats and conversations.”
BH: Automated chat flows and things it's the way everybody did it really didn't they before AI came along?
YR: Yeah, it was mostly about having a different user experience. So, having people looking up your website and instead of browsing around your product, having them chat with a representative asking them a few questions like “what do you want to buy? what are you trying to look for? What are you looking for? Under the hood it was basically just a decision tree that just helps them browse better on the website and chatting all day, and we had that running for a few years, we raised money. Actually, the product still works to this day, but we don't have like one customer that's using it and we someone that is still operating it. I moved on to an actual generated AI start-up which was Tabnine.
BH: So, tell me a little bit about Tabnine. So, what did they do and what was it about what they did and the kind of the AI element of it that grabbed you?
YR: What grabbed me was that when I was connected with them, the first thing I did was go through their website. Their previous name was called ODA. I know I'll tell that story in a minute. So, I went onto the website, and I immediately understood what they did. Now the product was for - is - for software developers and I don't know anything about coding. I never wrote a line of code in my life.
But I understood it right from the…just by reading the headline it said - I don't remember exactly what it said – but it was basically an “AI code completion” and I understood “OK so people are writing code and as they type it automatically generates the code for them.” That really attracted me because I was already invested in the field of AI and completion or generative…back then nobody used this term generative AI so that really attracted me then I joined.
Tabnine was actually a one-man show there was acquired by CodeODA, and it was back…and now we have ChatGPT four…it was back when it was just GPT or maybe GPT2 and it was open source. So, there were many people - mostly hackers, the good type of hackers - they were kind of playing and it matured into a real company where they said, “OK now is the time.” before that there was not enough code available in the world that we can analyse. Machine learning wasn't good enough back then. It was already at a level that you can train models on millions of lines of code and then become a product that knows how to complete the code for you in the same thing we already had in Gmail and Docs you know like you type an e-mail and suggest some words for you and sentences...
BH: I mean I'm just looking at their website now it's just it's very clear. It's an AI code assistant that makes you a better developer.
BH: In exactly the same way that if you kind of use ChatGPT or any of the other sort of generative AI technologies and you use it correctly and you know the right questions to ask and you can have the right starting point. It just makes life so much easier.
YR: It does, it does, and it’s amazing.
BH: It’s very cool.
So, you want to Tabnine as their, CMO, Head of Marketing, and built a team there.
BH: Tell me what happened next?
YR: When I got there, in the market it was Tabnine, and there was another company, a start-up company, that pretty much raised the same amount of money like 10 or $20 million from the US. Their name was Kite, and the promise to the market, and to investors obviously and everyone, was “it's only getting started.” OK but at one point everyone in the world, the 10s of millions of developers in the world, will be writing code with AI. I mean there's no other way out of it.
And we grew really fast, and everything was all good and the fun and we were beating the competition. I should say maybe the other competitors that we had was AI that wasn't really AI, that already existed in the environments of developers. OK so you had to pitch it to developers, and it was a more of a B2C company, so we didn't really yet pitch to developers. But the promise was “sure you have like a poor imitation of AI completing your code but we can do much better than we can generate much larger chunks of code for you or lines or whole chunks of code.” And then we grew really fast, and everything was going pretty OK. But the market wasn't yet big enough.
Because when we did end up talking to VP of R&D you know, CTO, like Chief Technology Officers. So, technology managers and company. When we wanted to sell them the enterprise product and we will tell them “Look everyone's going to do any anyway so why not help your team become much more productive and cut off 30, 40 sometimes 50% of the time that they spend writing code?” and people were like “okay it looks nice but yeah, I don't know. I want my team to write the code by themselves” [there were] many different excuses.
BH: Why do you think people would have…is that this idea, do you think, that generative AI, or AI more generally, is somehow cheating if you like. People need to be able to do it themselves?
YR: That’s the thing that…I didn't know it back then because, again, I didn't know anything about coding, but developers were already spending a lot of their time Googling for code and then going to Stack Overflow, or somewhere else, and copying the code into their ID’s, ID is the integrated developer environment, so. Anyways, part of coding was already copying and pasting code, this way or another. Everybody knew it. It's not it was a secret you know, even managers [knew]. Otherwise, you could never explain the success of websites like Stack Overflow for those who know it.
But still there was some resistance to it and it was a very hard position to be in where you are convinced that this is where the world is going to. But you have the decision-makers keeping the door closed and saying, “I don't know.” But then Microsoft came and that's where the story gets much more interesting.
Slightly before Microsoft got involved, we heard that kite.com, our competitor, was shutting down and we were really bummed up. Because the message to the market was “this AI for code thing is not going anywhere/” I was speaking directly to the investors, but I imagine that if I had to do it those calls with the CEO were probably hard cause the whole idea was to promise them that “look this is going to explode.”
I personally had a pretty rough couple of weeks with myself and then trying to motivate the team that “don't worry you know it's still going to happen. Everyone is going to be doing it at some point” and then two weeks later, out of nowhere, GitHub which is…
BH: Cause Microsoft owns GitHub, right?
YR: Yeah. Microsoft bought GitHub for $8 billion a few years ago. So, GitHub announced GitHub Co-Pilots you can already learn from the name “Co-Pilot what it means.
YR: So, they loved their AI code-assisted generative AI, whatever code completion whatever you want to call it, they launched a direct competitor to Tabnine. I was shocked. Everyone was shocked. But then I remember 2 hours later I became so excited about it because I said, “OK now we have the biggest company in the world.” Even Satya Nadella, the CEO of Microsoft, tweeted about it and wrote “This is the biggest thing that will happen in the next 10 years” and I remember writing a long message to my team in Slack pumping them up with motivation saying “okay so now we have a decent rival that's going to take the market by storm but we're going to be with them doing this” and then everything changed.
In one day, I think Microsoft and GitHub Co-Pilot managed to educate the markets and saying “OK you've been writing code so far in one way. But now it's changing. Now you're going to have an AI assistant writing the code for you in many instances” and then all the resistance that we face was suddenly…it just changed completely.
BH: So instead of your marketing really being about “there's a new category. There's a new way of doing things, you know, we’re the leaders in this category”, you had to kind of shift completely to a different strategy which is saying “well you already know that AI is transforming the category. But our solution is better because XYZ of a real kind of a challenger position rather than an innovator position I suppose in a way?
YR: Yeah, and I think that the messaging and the narrative about “who's being better or not” was very complex. So, Microsoft - it should be noted - had invested $1 billion into open AI. OK? So, you have like a triangle here of companies.
You have Microsoft owning GitHub. GitHub launching the product, and open AI is the provider of that technology. Which is, in a way, you can say that it's pretty much owned by Microsoft okay? Because they not only injected $1 billion, they have the first rights of anything that's being developed.
BH: Of course.
YR: ...and I think they can even veto the usage of some solutions being publicly released. So anyways so they launched GPT3 whereas Tabnine was using GPT2. So, the argument was rough but then I had a lot of different arguments because they released it slightly too soon and then people saw that there's some code leakage and inaccuracies…
BH: Sorry, what's code leakage?
YR: So, the AI that was completing the code for developers using GitHub Co-Pilot was using code that had some policy issues or malware…
BH: Oh, OK.
YR: …and then a lot of accusations came around saying that “they're stealing code basically from repositories that were not allowing anyone to use the code.” I don't want to make this conversation mostly about all that but in general the story was that in one day the market has transitioned from being a nice, cute niche to the leading example of AI and using AI in the world.
BH: It's interesting isn’t it because you’ve essentially got this sort of great groundswell of something that's kind of taking over a whole area and there's a whole community, in terms of the developer community, that's talking about this at using this and yet the rest of the world is kind of carrying on oblivious.
It's only really in the last few months, I suppose you know, with ChatGPT3 and now ChatGPT4 and Bar and so on that the rest of the world has caught onto it in the same kind of way. But for you as a marketeer working in a business where there’s been such a shift you know what kind of strategies did you use? I’m just thinking about my audience we've got quite a lot of CMOs, what do you do when the category shifts in that way in terms of your marketing strategy?
YR: Yeah so, I mean it was a really major shift. It turned from educating the markets to speaking to an educated market. But then claiming why your product should be used and why is it better? Why is it safer? when you're competing Microsoft.
So budget wise I couldn't really compete with it right? There was no point we were trying so it's mostly about working with the community and having, you know, developer advocates and thought leadership influencers to use the product and speak on our behalf. Everyone is using Microsoft this way or another right? But the story was about convincing them that, even so, it still is safer and possibly even better to use Tabnine and not get Co-Pilot, but we knew that it's still far better to be #2 in a huge growing market, than being number one in a very small niche market.
BH: What do you mean by “better”?
YR: Well, I mean let's say you can be Pepsi Cola and not Coca-Cola. Maybe Pepsi’s second. Or you can be XYZ beverage that maybe people buy it in a couple of countries.
BH: I mean I'll take that! That’s a huge opportunity you know? You could look at the negatives and say well, you know, “Microsoft is coming and just kind of stolen this from under our feet. We we've lost a lot of the control we had but at the same time we've also gained this massive opportunity. You just have to be able to be agile and if you stick to your existing, kind of brand purpose which is about ‘making developers’ life easier’.” It’s actually not going to work because somebody else is saying that kind of loud and proud.
You need to switch that messaging to really focus on, as you’ve said several times, like the ‘safety’ aspect of it which is just a small subset of what the category is all about so it's a real it is the messaging actually that has to shift more than anything.
YR: Yeah, and I took inspiration from Avis, you know the rental car company?
YR: For years their slogan was “we try harder”. Which was a really humble way to say “we don't have the budgets. We don't have the ability to be number one in the sense that we're going to be all over the place but we're going to try harder.”
BH: We're going to talk a little bit more about generative AI and see how you're using it as a marketeer now so before we do that, you're obviously no longer with Tabnine, you are VP of Marketing at Brew. So, tell the listeners a little bit about Brew and what Brew does. Because we use Brew in our agency. I think it's fantastic tool, all around kind of share of voice and building an understanding of the conversations that are going on. But tell me a little bit about you and what you're doing now.
YR: Sure, so Brew, in essence, what it does is it just looks at the entire web and the domain that you, as a customer is working in. You will provide us with a few prompts like you will say “okay this is the target audience I want to go after. These are the topics that I want to track.”
You just have to give a few, like three to five, you know, a handful of them and then we go out and we expand, we have our own AI that looks at the entire web [it] understands identifies all the marketing activities in the market. We gather them and then bring you an entire web application full of dashboards and reports and analysis that you can then go and understand what's happening in your market, understand what is your share of voice in different segments and environments, how well is your engagement comparing to others on social media and other channels, and we also have the ability to provide you with an actual marketing plan based on whatever inputs that you give us. So, you'll give a few inputs and they we’ll say “okay that's where you want to achieve? Great, here is the marketing plan that you need to do.
So we have a lot of marketing teams and companies working with Brew, you know, planning their strategies, launching products with Brew, to understand how the market looks at before, understand which events they need to attend because we know, we show them which events have the most engagement, and I will say relevancy to their brand, and this really has a lot of turns that you can take using Brew which is what I really like about it.
BH: Yeah so, we worked with Brew on a project recently around unified comms, and actually I think for us as an agency, it was quite nice as we can provide that conduit and just help manage all the inputs into Brew as well which worked really well for clients as well. But just to give listeners a real example. We were looking at an integrated communications provider that was really trying to compete with some of the household names. The Zooms, the Teams and some of the other brands that were much more well-known and actually the tool was really great for understanding not just what channels people were using, but what they were talking about, and the topics and I think that was what was most interesting for us was around the messaging.
Cause it’s very easy to just, kind of, follow the same topics everybody else's talking [about]. Everybody's talking about hybrid comms. Everybody's talking about connectivity. Everybody's talking about generative AI, or whatever it is, and I think you know as a marketeer, as a brand, I think, you’ve got to, you know, you have a choice to make.
You can either come up with a strategy which says, “I'm going to talk about the same things but I'm going to do it in a different way, we're going to be part of that conversation”. Or actually, and I think, you know generally speaking, the smarter way of doing things is looking for those gaps. What are the things that, you know, your customers, your audience that the influencers in your space, are talking about, and actually your competitors are not, and I'm not sure there's an easy way of finding that information other than using something like Brew. The hours and hours of desk work you’d have to do for that and even then, it would be a little bit finger in the air, I think.
YR: No, no you’re totally right. I mean there's really not a scalable way that you can actually, not only find information, but also understand it. So one of the things that we do we use NLP, a natural language processing technology, we have researchers that, they find the context and the syntax and they understand the connection between many different activities to enable use of… just to get that up in front [of you] and show you “OK this is this is where you could have been” or just “where competitors are succeeding and you're not” or maybe vice versa.
BH: Bringing us back to the conversation of generative AI. There's been lots of stuff in the media just over the last couple of days around the dangers of AI, you know, Elon Musk who actually Co-founded OpenAI, you know, saying that AI is one of the biggest risks to civilization today.
Do you think it’s going to run out of control? Do you think that's what's going to happen? It feels a little bit to me like trying to regulate it now is a little bit like closing the stable door after the horse has bolted.
YR: Well, look I'm not a prophet so I can only speak on my own experience. But I don't see the danger in that way. I can relate to someone saying “let's continue releasing all these amazing features and new technologies with AI” almost on a daily basis. But maybe instead of publicly releasing it to everyone, let's only reduce it to a subset of users and see what happens, right?
So, I can relate to that approach, and by the way, it's already being done this way like Google's Bard OK? It’s not available in Israel. I can't use it. It's only available in the US, maybe Canada I'm not sure. So, they're doing it very carefully which is very Google right? We consider differences between OpenAI being fully out there, you know, now they're starting to monitor themselves more carefully.
BH: Yeah, I'm on the wait list for Bard, I haven't got it yet. I'm on the list, waiting patiently.
YR: There you go, and you’re based in the UK not some small country in the Middle East, right? So yeah, I think so it's just different approaches to it and, look, I don't know what's going to happen in 10, 20 years. I've no idea. But, on one hand, I think I want it's good to say, “OK there could be some dangers to it.”
But I mean, these days, the way that generative AI is being utilised, it's mostly for writing code, creating images, creating text. Obviously for medical processes there's already some solutions out there. But there could be things that are being used by generative AI that I'm not aware of.
My team are early adopters in nature so they’re using it. Like the other day My Growth Manager in my team, he told ChatGPT how to build a table based on different criteria in different columns and he’d been generating copy for ads, and I think he generated about 5 pages which was something that would take him probably the entire day, and he did it in less than 30 seconds.
BH: Yeah, I’ve been running some sessions for… initially with my team but also with clients around the different things you can do with ChatGPT. Cause, you know, there's a cynicism. A lot of people say “well you know it can’t write copy. It can’t write content in the same way as a human being.” Of course, it can't.
But there are other things that it can do. I mean one of the things I use it for regularly is to scan articles for me. So if there’s an article that’s, you know, or a piece of content that’s a couple of 1000 words long and I'm pressed for time, I can copy and paste it into ChatGPT4 and say “tell me what the four key points are in this article that a person that's running a marketing agency would find most interesting” and immediately I get the essence of it and I can then decide whether or not I want to read the article in full, you know, and just that time saving, you multiply that by days and months and years and I think every CMO can, you know, could benefit from that.
YR: Yeah, yeah. Of course, and then there's already so much content and guides and how-to videos and …like I've never seen in my entire life [so much content] about one topic. It's so quickly shifted from “how to use it” to “how to give the best prompts” and so it's always changing and they’re evolving.
BH: I guess, kind of coming back to the subject of this podcast or the audience for this podcast, I think, you know, there's a real strong message here, in terms of the kind of client-agency model as well where I think you know agencies that stick to the old way of doing things and charge clients by the hour, you know, and are still charging hundreds of pounds or dollars for creating articles that are written without the aid of generative AI are going to need to really think about what's their business model and how do they make money and how can they offer a better service to clients.
Because you should be able to use generative AI to speed up a lot of the manual tasks and to focus your time on creativity, on strategy, on adding value I suppose. On added value tasks, and I think those that are not able to do that, that they don’t have the capability to do that are going to find themselves in quite a tricky situation.
YR: I agree most definitely.
BH: It's been brilliant talking to you. Do you have any, kind of, final thoughts or things you wanted to kind of share with the audience in terms of, particularly around, kind of, either generative AI or around this kind of shifting markets when you suddenly become a challenger in a big category rather than a big player in a niche. What might your final thoughts be?
YR: Yeah, good question. I didn't prepare for it! But I can still…
BH: I like to catch people out!
YR: Yeah, probably should have. I think, I think there's two main events that are colliding right now in the world which is, on one hand there's an economic slowdown, you know? A lot of companies are, budgets in general, there's massive layoffs. Anywhere. Big tech and others. Obviously marketing budgets are being cut-off. But at the same time, you have all these incredible solutions powered by AI that can make up for, I would say, some of those cut offs. So, I will say to marketers don't hesitate to use them and to carefully see which solutions that, you're still out to buy, have the existing potential but also the future potential, in the next few months, to become much, much better and, you know, just make up for all that loss of budgets and power. I think that's the only thing that I can think of right now you know if I had to say something.
BH: Now that completely resonates.
BH: The point about generative AI, I guess, is that it makes it easier to make intelligent decisions and intelligent analysis of what is essentially a huge amount of data and, I guess, one of the things that I'm excited about is not just for marketeers, but for business leaders particularly in businesses with purpose, about how it allows people to analyse scenarios and do new things in a way that can make a real difference. I'm particularly thinking about climate change, and about diversity, and about equity and about how businesses essentially become a force for good in the world, and actually, if you can use generative AI to look at different scenarios around, I don't know, crop planning or feeding populations or how your brand could develop. I think that's so, so powerful and it goes beyond just marketing as the…in terms of marketing comms, in terms of the voice, in terms of the messaging and the channels but really down to the purpose of the business as it stands.
YR: Yeah, I couldn't agree more.
BH: Thank you so, so much for your time. It has been absolutely fantastic speaking to you, and I wish you all the best for the future. So, to my listeners, if you want to find out more about Brew there'll be a link in the podcast description and there’ll be more information, as well, around kind of measuring share of voice, and how we use these tools, on the BH&P website but for now I just want to say thank you very, very much to Yoav for being my guest today on the show. I hope that I will speak to you again soon.
YR: Thank you. Me too. Definitely looking forward to it!