AI will become the #1 inbound channel | Todd Sawicki, Gumshoe AI

Trailer Voiceover:
Todd is the founder and CEO of Gumshoe. Gumshoe has built a platform that helps brands understand what AI thinks about them and then what to do about it. How should AI companies and founders, how should they think about customer acquisition in this new world? Think of your email inbox and how flooded it gets, right? Automation of outbound has made customer acquisition through those sort of channels as hard as ever. One thing I've understood is that every successful startup has figured out at least one channel which really started working for them. I think in today's world if you're not figuring out AI visibility and AI discoverability, I think you're going to miss out. Especially in enterprise, we're certainly hearing about the number of startups that are seeing AI as their number one inbound channel is very high. So what's happening today is, if you think about it, if I produce information, it's consumed by the models and within 86 days it's going to regurgitate that back out to other users. Our point of view is that in 12, 24, maybe 36 months, that's going to be single-digit days. Tomorrow that could be almost real-time conversation. And I know that you also worked with a venture builder, PSL, which they have been around. How was that experience? What can you tell us about your experience? Why did you decide to work with a venture builder? Push the limit.

(Music)

Amit: Hi Todd, thanks for joining the Build AI podcast, and we're really looking forward to talking to you. You know, for the audience, you are the founder and CEO of Gumshoe, and it will be great to know a little bit about your background to get started.

Todd: Sure. Gumshoe is a startup that has built a platform that helps brands understand what AI thinks about them and then what to do about it. And as a larger background, this is, I don't know, the seventh or eighth venture-backed startup that I've done, always in the media or marketing tech landscape. And so it's been an interesting journey to see the rise of AI and, you know, be a part of that.

I've had the background... a long time ago started in the search space in the toolbar space as Google was launching. And then ended up being in the social space as Facebook was launching its platform. The original product managers for the Facebook platform, like Ben Lang and Josh Elman, were people that I know. Actually, the startup I was at was called Lookery, and it was actually one of the partners featured on the first F8 slot of partners in the space. Sadly, the landscape went through a lot of changes. The startup didn't quite work out the way that we'd hoped. And then ended up working for a meme publisher called Cheeseburger, as in Cheeseburger, the kind of the popularizer of LOLcats. So, if you get silly cat photos from your grandmother or mother, it's literally my fault. I'm one of the people who made that possible. So, I apologize in advance for that.

Then ended up going into the programmatic space and the paid space and helped build a startup there that was eventually acquired. And then was in the e-commerce space, kind of dealing with retail media. And now I'm off to the world of AI search. And partly what's just been interesting is how my career has gone from one channel to the next and building tech as that emerges. It's just been coincidence along the way. And now I focus on the problem of helping marketers understand AI and AI search and what to do about it. So, it's been a fun and interesting journey to see that evolution through the years.

Amit: Right, thank you so much, Todd. As any second or third-time entrepreneur will say, customer acquisition is kind of the most important thing. It's also now, in the AI world, the hardest thing because it's become easier and faster to build products. And, you know, as you explained, you have been doing marketing tech startups for like two decades. You went from search to social to paid, programmatic, to retail media. And you have seen the entire journey of how customer acquisition is happening throughout these phases. Right. So when you look at it today, right, which is like half of the companies are projecting themselves as AI companies and it's cheaper and faster—only half? Isn't everyone now an AI company? It is, it is, some people are catching up. But my question is that how should AI companies and founders, which is a lot of listeners, you know, of this podcast, how should they think about customer acquisition in this new world where actually there are a billion weekly active users on ChatGPT and, you know, Perplexity and Claude and so on and so forth? How should they think about it?

Todd: So that's a great question, and I think there's an interesting kind of perspective beyond that to start with, which is: I think customer acquisition today with the rise of AI and AI-powered tools is as difficult as it has ever been. And if you look at even if you're in the enterprise space and you're trying to do enterprise sales and outbound sales and what have you, the automation of outbound has made, I think, customer acquisition through those sort of channels as hard as ever. Like think of your email inbox and how flooded it gets, right? You're nodding your head at me like, yes. We've gotten to the point where it's almost impossible to get seen because everyone's inbox is so flooded. The cost of generating an email is now like literally zero, even highly customized emails, because AI can do that, right?

Before that was kind of, you know, with the rise of Outreach and SalesLoft and other tools like that in helping salespeople customize emails very quickly, that was seen as an advantage. But with AI, every email can be customized. And so suddenly... I think it's fascinating. I was joking at a dinner with other founders recently. It's almost like the best salespeople now have to be like it's catfishing. You have to figure out, how do you get someone to pay attention to you and respond to you? It is so much harder than it ever was because communication has been entirely automated. So in a world where customer acquisition is harder than ever, what do you do and how do you deal with that? It's an interesting problem.

And I think with the rise of AI search, there's an interesting opportunity to get discovered in an organic way and in a way that was almost lost as SEO got gamed and it was almost impossible to get ranked in Google because every keyword had been gamed to death and every keyword was just "top 10 this" or "top 10 that".

In a lot of ways, what's fascinating about AI search is—and I like to think about AI as solving the problem of traditional search. And what I mean by that is, one of the things that we do with Gumshoe is, you know, we're helping brands understand what AI says about them. Help them understand the types of people that AI is responding to, the types of prompts that those types of users are generating. So you kind of understand what AI is saying to those types of users and those types of prompts. And through that analysis, I've been fortunate in that the startup I'm building, I've got a problem where you know brands are worried—every brand out there is worried about what AI is saying about them—and I'm getting 100% of my users are organic and inbound. Maybe I'm biased in terms of the way to think about customer acquisition in the world of AI.

And part of that has been: if you can figure out the sort of this almost infinite longtail of search that's now emerging with AI—because you can ask anything—where with Google the most generic keyword you could make was always the best way to search for things, the more specific you are with AI, the better off it is. And the benefit of... we've got almost 6,000 companies that have signed up for Gumshoe in only like seven or eight months. And what we've learned through their activity is, as they sort of have run this analysis and create these prompts and understand what AI is saying to them, we have a good sense of what AI is doing and how it's responding and tracking the references it gives, the URLs. And through that analysis, what we've determined is that between 50 and 90% of the references—the URLs that the models cite—are traditional longtail SEO, meaning they'd be on page three or beyond in a Google ranking. And so in traditional search, like one out of a hundred people goes to page two on Google and one out of a thousand goes to page three and nobody goes to page four. And so what this is showing you is that the AI, as a machine, has infinite patience. And so yes, it's crawling Google links for discovery. It's using Google as a mechanism to help understand URLs, but it's reading all 100,000 or million or 10 million links under a search. And then what it's helping do is figure out what is the best response.

And so as a marketer, if you were really good at explaining your products and having well, authoritatively described products and things that are getting mentioned in forums like Reddit and Quora, you have a chance to get discovered in a way that is almost impossible in traditional Google. And so that's what's fascinating about AI, is AI search is solving the fact that Google is so gamed that it's no longer showing the best answer. It's showing the most popular answer in terms of like backlinks. Basically, think of links and the way Google does that is just a popularity contest of votes. Whereas AI can read all million URLs and help you discover the right one. So if you're the right product for the right person under the right set of circumstances, you can get discovered again. So organic is in a way coming back. And I think that's one of the lessons, you know, with AI search: suddenly you can get discovered in a way that became harder and harder and more impossible with traditional search. So a lot of the world has relied on paid for the last five, six, seven, eight years, and I think we can maybe see a shift back to organic as a result of the rise of AI and AI search.

Amit: Yeah, I totally agree. Actually, that example you gave about emails hits hard because I think with AI people are sending a gazillion emails and it's an infinite amount. It's unbelievable.

Todd: Yeah. Right. And I have another story along that line. Another great growth hack, right, that we all as startups do and SaaS companies do is, you know, we all send lifecycle emails, trigger emails. And as a startup, one of the most important things as you're building a product, trying to get to product-market fit, is you want customer feedback. So what do you do? You send emails. And the reality is, no one responds to those anymore. And so one of the fascinating things that I started is, my team was struggling, like "Todd, we're trying to send emails out. I want to get customer feedback." I'm like, I don't know how to get anyone to respond anymore. And so one of the things I did as a hack is—we have, you know, I've mentioned over 6,000 users have signed up for Gumshoe—I've literally now what I do is, and I'm doing this manually, this is not an AI, this is not automation. I now connect with anyone who signs up for Gumshoe on LinkedIn and basically say, "Hey, I'm the founder. Thanks for signing up. Not trying to sell them. Not trying to upsell them. They sign up for a free trial. That's fine, right? I'll still connect with them." And just to a point where I get feedback and that's been great because the amount of that channel now is probably our number one resource for customer feedback because people respond because I'm like a person.

So to your point about AI generating infinite content, we've had to figure out other ways of getting users to give feedback. And so I'd say for anyone who is building a startup, connect with all your users as much as you can outside of email because to your point exactly, I don't know how you get anyone to respond to email anymore. It's unbelievable. It's like if someone responds to an email, it's like you've won the lottery.

Amit: Yeah. And, you know, Todd, you have done many startups. I have done a few and invested in about 32 of them. One thing I've understood is that every successful startup has figured out at least one channel which really started working for them. You know, for Mailchimp, it was the podcast. For us at Medici, we had like 6-7 million page views and 100,000 people on the newsletter. So a lot of inbound. And a lot of people use paid. And one with it. I think in today's world, if you're not figuring out AI visibility and AI discoverability, I think you're going to miss out. So this is a very, very important topic and maybe because it is new, right now there's an opportunity in 2025, 2026 to actually sort of figure out this channel and make it your best sort of customer acquisition channel.

Todd: I agree. There's an opportunity today. Right. I think that's why we're seeing the popularity of—like I'm getting all this inbound interest—is the marketers who understand that. Right. I agree. Every startup figures out a channel that it can win with and it can dominate. Your Mailchimp example is a great one. And so I think it's the same thing, which is AI is an opportunity to get discovered and have it be a channel. And we're especially in enterprise, we're certainly hearing about the number of startups that are seeing AI as their number one inbound channel is very high. So that is a huge opportunity. And by the way, there will be a paid version of AI in terms of ads or something and that'll be a new channel. And so sometimes it's organic and sometimes it's paid. But I totally agree with you that you have to figure out your channel. If I figured out like LinkedIn chat is a good one for me, great. In addition to AI and AI search, obviously we're eating our own dog food there. But I think that's another thing: find something that works for you. I think it's great advice. I absolutely agree with you.

Amit: Yeah. And now let me shift the gears towards marketers and talk about so there's, you know, probably like a million marketers around the world who spend a lot of time, energy, and money to learn SEO and a bunch of other channels, right? And probably some of them are listening to this. And also in general, figuring out that the ground is shifting, right? And they probably are wondering like, where to start? And you know, how to—there will be a mindset shift required here. And I think the one thing that you could really help the audience with is: how should they start thinking about AEO or GEO? Are there like the first four or five steps that you need to take, you know, to get there?

Todd: So that's a great question: where do you get started? And I think one good thing is that a lot of SEO practitioners and marketers are now talking about AI. Whether you call it GEO—Generative Engine Optimization—or AEO—Answer Engine Optimization—or AIO as an AI optimization, or AI SEO, or LLMO, or whatever it is, right? We don't have a universal acronym at this point in time. And as a former paid media guy, GEO, like we have—it sounds like geo-targeting, so it's not been my favorite either. Just as an aside. But I will say, it isn't that hard to get started.

The good news is, you can—everyone can certainly go to Gumshoe. It's, you don't need a credit card. It's free to sign up. You can put in your company and kind of as a starting point. And our point of view is there's kind of three things that you're trying to understand from the point of view of what does AI think about your business.

The first is AI is personalizing its answers. So one of the things what that means is, as a marketer, you can't just log in to your ChatGPT account and get an honest answer for what AI thinks about your business. Because the minute you log in and put your email into ChatGPT, it's building a profile of you. It's using your email, looking you up on the internet, looking you up on LinkedIn and search engines, and it knows where you work, right? I mean, if you were to research any of your portfolio companies through your ChatGPT account, it's going to flatter you because that's how the models are designed. They're designed to please you. And so what that means is it kind of lies to you. It flatters you. So it'll say all your investments are doing great and they're the best in their category and what have you. So the problem as a marketer is—kind of where I think you're going is, well, Todd, marketers don't care what it says to them. It cares what it says to its customer. So that's kind of step one: you can't rely on using your own personal ChatGPT account or Perplexity account or Anthropic account to get an honest answer.

So then the problem is, well, crap, how do I get in the shoes of my customers? And our point of view at Gumshoe is that we believe you have to kind of understand that AI—the difference with AI search versus traditional search—is AI is absolutely customizing its answers based upon your profile. So it learns about you. It understands, like, are you interested in—and I think the best analogy is in terms of this and what you're interested in or not. You imagine you walk into Dick's Sporting Goods and you want to buy a new pair of shoes. And we've all seen that wall of shoes behind people. And as you go up to the salesperson and say, "New pair of shoes," they don't just grab a random pair of shoes and give it to you. They ask, "Well, are you—what sports do you like? Or, you know, if you're a tennis player, are you an aggressive player or a beginner? Are you a runner? Are you a marathoner or are you a casual jogger? Are you a hiker, like a backcountry hiker, or are you just hiking through Central Park in New York City?" Depending upon that, you will get a different pair of shoes, right? And AI search is like that salesperson at Dick's Sporting Goods. It reads the internet and uses that to kind of understand, "Oh, right, depending on who you are, I need to give you a proper answer." And so we've got lots of data that shows it really does vary its answers based upon the profiled user.

So number one is understand what AI thinks is your customer set, right? And Gumshoe as a platform will showcase this. And then the next thing you're trying to understand is, okay, what is it saying back to those customers? What are the types of prompts that are happening? And it's no longer keywords, it's now prompts. And trying to understand the prompts that are associated with different types of these personas or customers.

And then from there, what you're trying to do is—there's three things you can then do. Once you understand what AI thinks about you is kind of a level set: Is it talking about you or your competitors? And then within that, is it saying good things or bad things? Or what are the features and areas that it's focusing on? Then you can kind of do three things.

You want to make sure your website is as AI-crawlable as possible. And so there's some technical things that, again, platforms like ours will help you analyze your site to make sure it's as crawlable as possible. Because it turns out—and there's some news today, right, that came out about ChatGPT, how it's no longer going to allow users to pick the models that it's responding with—because it turns out some models are really expensive for ChatGPT to run and others are really cheap, and it's going to limit your ability to get the really expensive answer. And so if your website is difficult to crawl, it's expensive for the model. And so we have lots of data where we've seen sites get penalized compared to their competitors up to 80% less visibility in AI as a result of their sites being hard to crawl. So make sure your site is as crawlable as possible, right? As easy as possible. Takes as little effort as possible for AI to understand what you're doing and saying and so forth. So that's one.

Two, the next thing you want to do is you want to produce as much—what I like to say—authoritative information about your company as possible. So the level of—if you think about the potential universe of information you can be talking about about your products and whatever, most people and companies have focused on kind of the core 20%. Because as humans, right, the 80/20 rule, where you know 20% of the content is what Google cares about. It's what most users care about. It's not what AI cares about. AI wants the longtail of answers. It wants that whole constellation, right, of it. And so you're going to have to almost be as encyclopedic as possible.

The actual GEO term comes from a research paper produced by a professor and collaborators out of Princeton, and it actually dates back to late 2023. And he's the one who invented the term "Generative Engine Optimization." And in his analysis in this paper, kind of the core of what a lot of us are doing in the space—if you read that, what he talks about is that AI is a voracious consumer of information. Go back to my Google—or it'll read a million links, right? And so the reality is most brands only cover kind of the core information. Because, by the way, previously, producing content was artisanal, it was expensive. But with the rise of AI, generative content is inexpensive. So you have to cover a lot more information, because AI turns out—if it doesn't, if you don't give it the information it's looking for about your product or category, it doesn't say it doesn't know, it lies. It makes things up. It hallucinates. And so the last thing you want to do is make sure you're feeding an almost encyclopedic amount of information about your product and company.

So everyone—like the biggest blogs we probably see from companies—you might have 200 blog posts that go back four or five years. You're probably going to have to produce 10 to 20x that amount of content to cover everything that AI thinks is related to your business over the next 3, 6, 12, 18, 24 months. So the amount of content you're going to have to generate is going to have to be exponential because you want to fill in all those gaps. Think of—your job now is to be kind of the Wikipedia for your category.

Amit: Right. I'm sorry. I know that you have multiple steps. I don't want to break the flow, but this is a question I have always wanted to ask. So are you saying that you basically put out a lot of information, publish like hundreds or thousands of blogs, and write a lot about your products, not less? And then another question in that is: most of the people will then use AI, ChatGPT, to actually write that content. And then does it help or does it actually hurt? Should you be humanizing it? Should you be writing a lot of this content yourself? Can you clarify, because a lot of people have this question?

Todd: That is a great—no, no, it's a perfectly good, and by the way, happy to have you jump in and ask this as a follow-up, because I think that's actually a really raging discussion. For traditional SEO, Google was like, "Don't produce a lot of content." Well, that's because Google, in the world of SEO, crawling was expensive and it couldn't analyze as much content. It was basically saying, "Look, we're overwhelmed. We can't do it all." Well, AI has changed that now. So you want to produce levels of content beyond what you thought was reasonable before.

Now the follow-up to that is: you want to produce informative content, right? You want to go deep. It's not AI slop. It's not just generating crap. It is—you want to be detailed. You want to be informative. And now it's almost like I want to answer every possible question, every permutation of a question that could be imagined. And so that's, I think, if you think about it, it's like think about your—what I like to say is you go back through, if you've been around for a while and you look at your customer service logs, you're going to find thousands of questions that weren't answered by your website. They got answered through a chat, right? Because you didn't have time and whatever. Well, now you want to turn that around and publish all those questions and answers. You want to publish all that information, right, to the level of detail that you never thought people would ever care about. Well, AI crawlers care about it, and AI cares about it because it can consume all that and has the ability.

So I think our point of view—and we have data to back this up, and some of the members of our team are in the PhD program at the University of Washington, their AI program. And so we base a lot of what we do on academic research. And when you look at the academic research around this, the academic research says no, the AI engines want more information because they're veracious, they want to learn, they want to know. Now, it has to be factual. It needs to be cited. It needs to have good, legitimate information, ideally quote unquote "authored"—I'm putting air quotes around that—by a human with expert knowledge and expertise in the space. So they have authority, knowledge authority in the space.

The other thing about AI is AI doesn't care as much about domain authority like traditional search does. It cares about knowledge authority. So what it means is that person doesn't have to actually have written every word, but what AI really does like is it wants someone with knowledge and expertise to have reviewed it, to kind of signed off on it, right? And so that's, I think, the difference here. If you're producing content at huge scales, but it's good, legitimate, knowledgeable, factual-based, informed content—like put references. If you say something, reference a fact, reference another piece of information and so forth—then AI seems to really like that. And so it's you lying about things or misleading things is more of a problem. But being more factual-based and you can produce lots of this commentary, then and sincere. I think that's sort of the angle that our point of view is that if it's—if there's lots of it and there's lots of permutations of it, like, "Oh, maybe I want to answer it slightly differently for this type of persona versus this other one or this other audience segment," yes, it's good to tweak that. Traditional SEO didn't like that. It couldn't understand those differences very well. But AI is different. It can, and it's trying to. AI is expensive—like those models are expensive to create because they're processing an incredible amount of information. They're really trying to understand semantically what's being discussed. And so I think that's part of the differences.

So our point of view is yes, producing again quality information, human-reviewed—AI loves human-reviewed, right? So you could put a name on it. The other thing is that with AI, and this is where it gets really crazy, is there's this incredible recency bias with regard to information. So going back to my comment earlier, AI wants to put the most canonically correct piece of information out, whether it's the millionth link in Google or the number one link. And so to that point, if you think about it—which is, what's the most canonically correct piece of information, the fact that was published yesterday or 12 months ago? The one that was published yesterday. So our data shows that over 50% of the content cited by AI was published in the last 90 days. And almost 90-plus percent was published in the last year. That percentage published in the last 90 days is declining 10 to 15% quarter over quarter. So the average age of a piece of content cited by the models—publish date, right, to today—is 86 days old, and it's falling 10 to 15% quarter over quarter. And so effectively, what's happening is the models are trying to give the most recent version of an answer because it's probably the most up-to-date and the most likely correct.

So what's happening today is, if you think about it, if I produce information, it's consumed by the models and within 86 days it's going to regurgitate that back out to other users. Our point of view is that in 12, 24, maybe 36 months, that's going to be single-digit days. And so the way that we like to think about it as a marketer: that content you're producing, think of it like a conversation. Like you're training, like you would if you were a brand selling into Dick's Sporting Goods. Reebok comes in every quarter and they do a training session with all those salespeople in every store, and they're basically saying, "Look, here's the latest products. Here's the features and benefits. Here's why we're better than our competitors." And the salespeople hear that and they use that as part of their day-to-day selling to the end customer. And if you think about the content you're producing now as a marketer, it's kind of the same thing. You're having to train the models. The models want to be trained. They want to represent your products well. They want to answer questions well to their searchers. And so that's today: an 86-day cycle. You train them, and 86 days later, the responses.

But in 12, 24, 36 months, that's going to be single-digit days. And you're basically going to be having a conversation with the model by producing content. The model's going to consume that. It's going to respit that out. It's going to send traffic to your site. And then basically, you're going to have to respond and say, "You sent the—oh yeah, thanks models, you sent the right traffic," or "No, models, no, no, no, change it. You want to talk about me like this," because the models want to send traffic that converts, because that means that they're giving the right information to potential customers versus wrong information. And I think this radical change, where today I can publish content and it might be evergreen—that's going away. And we think you're going to be now in this rat race of talking to the models, meaning producing content to speak to them, and then responding to what they send to you. And again, today that's an 86-day conversation. Tomorrow that could be an almost real-time conversation, which is kind of scary as a marketer. Like, "Holy crap, I'm going to have real-time marketing is coming." And it's going to be fascinating. And so I think this change is going to be just amazing to watch.

Amit: Right, so Todd, help me help me summarize what you said. Right. So you spoke about in the beginning that ChatGPT and most of the AI flatters you, you know, and tells you things which you are directionally already asking and will tell you good things about your companies or your products. So you have to take a view from a customer point of view: what is the persona and how they may be searching and what kind of results they may be getting? So you can't use your own account to do this. You should probably use either like a Gumshoe type of a thing or like a third-party account. You said you have to make your websites AI-crawlable, and you sort of design and do things like that. And then you said put more content, put a lot of great content answering questions that your customers might be asking. And what else, if you can just help.

Todd: Well, the last thing that you want to do is—AI does care about authoritative mentions. And so there's going to be experts out there that it references. And so you want to make sure you're getting mentioned by them. And so part of what's happening is we're analyzing the URLs that are getting mentioned or cited as the rationale for why they answered the question the way it does. And so sometimes it's, you know, 30 to 40% of the time it's the brand's own content. The 60 to 70% of the time it's other third-party references. So you're going to want to be mentioned in communities like Reddit and Quora, or in blogs—like traditional PR and outreach. And you're still going to want to get those mentions. So that's the last thing: as you discover people who are covering your space—and what's interesting about AI is, again, AI, unlike traditional Google, only the most popular sites are getting referenced by Google. So what that means is every company in your space is reaching out to the exact same five or ten people, and they're just getting overwhelmed with inbound, like, "Oh, right about me, right about me, right about me." Like, we've all done it and you've seen that.

But what's fascinating about AI is, because it will go dig through and find these longtail experts, you have opportunities to do outreach to people who usually aren't getting discovered because they're on page 100 of Google's results. So that means they get one click a year, but AI goes, "Oh, this person really knows." Maybe it's a professor. Maybe it's a real deep-in-the-weeds category geek about something. And their blog gets 20 people a year, but it's about the people who absolutely care about something. And those people AI loves, and you want to do outreach to them. So we're helping you—you want to discover who are those writers that AI likes to reference and making sure and finding a way to get them to talk about your brands as well. So some of the traditional marketing strategies: yes, you want to produce interesting, authoritative content, like your newsletter as a good example, and you're wanting to do outreach. But unlike today where you might do traditional outreach just to the most popular sites, well now it's what are the sites that are getting referenced by AI, which would be more knowledge authority sites—not necessarily the most popular sites. And I think that's one of the things that sort of arises too. So that's sort of: make your site as easy to crawl as possible, produce as much encyclopedic-like content as possible, and then you're also going to do more traditional outreach to make sure you're getting as many mentions by some of these knowledge authorities as possible as well.

Amit: Got it. That's very helpful. I want to sort of come back to the journey of how you sort of started Gumshoe. And I see you have a fascination on this field itself, right? Do you believe that AI visibility becomes like an earned channel, like SEO, or inevitably a paid channel like everything else?

Todd: So I think it's going to be like search, where it's going to be a blend of both. And where there's traffic, people will pay to get in front of the line. I mean, you look at Disney. If you want to get to the front of the line, they'll sell you a very expensive pass to cut the line, right? And I think this is just human nature where businesses will be like, "You know what? I want to be mentioned there, and I don't care what it costs me. I'm going to be there." And companies like OpenAI need money, right? They're losing billions of dollars. And so I do think there's going to be a paid option, but it's not going to be "you have to do paid." It'll be kind of an either/or, I think, kind of point of view. Yeah. Some companies will be willing to pay and others will be good with organic. And I think it'll depend on, as you said earlier, some companies have done extremely well with paid strategies and paid acquisition and have built amazing businesses, especially in the DTC space. And so I don't want to say that paid is—and I've built companies that, you know that DSP that helped companies buy ads—and so I think it just depends on who you are and the company and the circumstance.

So yes, I fundamentally believe that there will be paid options inside of the AI models. I mean, look, we're beginning to see it emerge. I was recently at a discussion where Google was talking about some of the new ad ideas it's testing—things that might never see the light of day, but they're absolutely thinking about paid options for AI and what that might look like. I think the first place you'll see paid show up is, you know, a company that—the company that I sold my DSP to, which is now known as TEYES. They're now building an ad format because they've always been in the PPC space. What's been popular actually is a lot of game companies and other fictional kind of content. What they've done is created chatbots around characters. So if I want to interact with a character from a game or from a book or from a movie, this has become a very popular thing. And they become paid—sometimes paid and sometimes not. But obviously, running those AI chatbots is expensive. And so now, these publishers are looking to put ads in these chats. And so I think that will be kind of the first experience where you see how do they contextualize those chats? PPC or text links make sense. And so I think we're going to see TEYES just launched a product to add an ad product for these chats for publishers. And so I think that's kind of the beginnings of it: how does—whether it's an AI search chat or a character chat or what have you—I think we're going to see the testing. And so the fact we're seeing some of these tests now emerge—again, TEYES just launched this I think last week—and I've heard rumblings of Google doing similar. And so I think that's where we'll see it. It'll be outside of AI search first, and there'll be AI-powered experiences like these chats, and I think those will be where you'll see that. So people should look—you kind of want to see where the innovation is going to occur. Look at these tertiary experiences, right? These tangential experiences, and I think we'll see the innovation there. What works there will eventually, I think, make it into AI search chats as well.

Amit: Fascinating. No, that's great. Um, thank you so much for taking us through this whole journey and you know how AI visibility is becoming important. Switching gears here. So I was actually doing some research on your profile and, you know, going through LinkedIn and I found it very fascinating. You say that you have been an investor in 500 Startups, Founders Co-op, Ascend Capital, Contrarian Capital. So these are like—you basically were an LP in these funds?

Todd: Yes, I was an LP.

Amit: Right. And then you have a lot of roles, you know, as you know, in the advisory board for startups and consultant. So tell us a little bit about the journey. And, you know, there's so many of them that you have helped or invested in. And then after doing all that, how do you again sort of start a company and go through the toil? And, you know, like when you're doing a startup, all the fires are burning, all the odds are stacked against you, and it's a lot of hard work. So how did you get back to sort of doing all this? And I have pretty much went through the same journey, so I'm very, very like fascinated to hear yours.

Todd: Well, I think what happens as a startup founder: you either become eventually an investor or you just keep starting companies. Because I think what happens is you become broken. Meaning you can't work in any environment other than startups. Starting companies—like prior to doing startups, I worked at a big—I worked at Disney and Price Waterhouse and things like that. So I've been working in companies. And I could not work at a big company. I'm just not built for that. And I think, you know, we all like—you've got the experience of startups. Startups are their own unique cultural beast. They're hard. They're insane. I have friends who work—I live in Seattle, so I have friends who work at Amazon or execs at Amazon or Microsoft—and they say, "Oh god, I'd love to do a startup. It sounds so awesome, idyllic, freed from the corporate shenanigans and Game of Thrones politics." And I'm like, "You people have no idea what it's like to work at a startup." And you have to be insane to work at a startup because no one takes out the trash. There's never enough people. You never know whether you're going to run out of money. The amount of stories of even really successful companies that couldn't raise money in their first rounds or struggled in their middle rounds—they were like on death's door. I think you have to be just—I tell these people, "You're insane. When you're ready to get paid peanuts and take out the trash yourself, then call me and let me know you're ready to work at a startup." Because I think you're right in that it's hard and it's painful and it's not built for everyone.

For sure. I think you really have to believe. And I think if you like the thrill of the zero-to-one stage, for me, it's the most amazing journey: going from an idea to an actual thing and offering, and then when you have customers pay you for something—holy crap, that's an amazingly exciting journey. And that's the rush that I get. And I'm broken. The politics of companies—like ultimately, I got a—I'm familiar with TEYES because I sold a company to them. And you know, I was only a thousand people, but after working there for a few years, I was like, "You know what? I'm less and less talking to customers and more and more doing internal meetings and discussions." And by the way, they're wonderful people who work there. I love my co-workers. But I love more working with customers and building things for customers and talking and engaging and helping them solve problems. And so that's what got me up every day. I realized that, you know what, I'm not an investor. I'm a builder. And I think that's fundamentally how I looked at the journey.

I'm better—I invest in other funds because, you know what, I'm better at letting other people be good at that. And partly, I think as a startup founder, you become very cynical. Like, "Oh, that's never going to work." You're actually—I think most startup people, people think, "Oh, you must be a blazing optimist." I said, "No, we're all cynics." We all see where the world's broken, and that's kind of where we get ideas to go build things. We're really good at figuring out what's broken in the world, and then we figure out a solution for it. And so as a result of having done many startups, I'm like, I'm too cynical. I think I can find you a hundred reasons why a startup will fail. Because by the way, I have to do that same rubric on myself for every idea I come up with, because otherwise you'd start and do too many things. And so I think I realized that I was best at building and letting other people be investors who are better at identifying that person or that idea is a good one. And I can see through the—how that could be a winner. And I'm just not a big company person because I prefer to build things and get the rush of selling things to customers and solving their problems versus just the machinations of corporate Game of Thrones life.

Amit: And I'm smiling here because I can 200% relate to everything that you said there personally. It hits personally basically because I actually went through the same journey. Built and sold a company, and after that I was thinking that maybe I should start a fund. And I spent almost 11 months talking to about 50-60 VCs and laying down all the foundation work to start a fund. And then I realized I was looking at like 30 companies every month and investing in a bunch of them through a syndicate and so on. And then I realized I really don't like doing this all the time because it's like a critic job. You're looking at decks and startups and then you are basically critiquing like, you know, this is good, this is bad, this is ugly. Then you invest, and then after investing you're doing the same thing. And I was missing the action of building and going through that pain and suffering that you described. And that's why we started the venture builder. Because the interesting thing about a venture builder is you get to do both: you get to build and go through that journey along with the founder, and then you also get to invest. So it was that journey there. And I know that you also worked with a venture builder, PSL, which they have been around. How was that experience? And how do you—what can you tell the founders? Because I feel like there are two types of founders. One who find this venture builder structure really really interesting and they want to work with them. And then there are others who feel like, "Oh, we don't want to give up this much equity to a venture builder." What can you tell us about your experience? Why did you decide to work with a venture builder?

Todd: So, it's a great question. And I'm working with Pioneer Square Labs. And part of that is the guy who started Pioneer Square Labs is a guy named Greg Gottesman, and he was an investor in one of my companies before in the traditional venture model. So he and Greg and I go back well before he started Pioneer Square Labs. And when, after my previous startup to this—which was in the retail media space—I was trying to figure out what to do. And what's—to me, when you're starting a company, it's very lonely. And if it's just you and potentially a co-founder and you're working out of coffee shops, it's hard. And what's nice about—you know, a venture builder, you know, like your own—is you have more structure and there's more resources and there's more of a community.

And, you know, in terms of—I think a lot of the times, you're in a position also sometimes financially. Like, I did an EIR role within Pioneer Square Labs because it was sort of like I wasn't quite sure of what I wanted to do. I had some ideas, and it gave me some flexibility to kind of explore, and some structure, and some community, and right, and that flexibility is like financial resources, in effect. Right? I think there's two models within the venture builder side, which is you know sort of the EIR role, which existed kind of inside normal funds, but that was always—those are always weird. Right? It's just kind of a square peg, round hole thing. And versus—and sometimes it's, you know, you as an investor have an idea and you're trying to find people to maybe run with it and build it. So sometimes you might—I can come up with an idea and I might hear some ideas from folks like yourself, and that's a good way of getting started. But don't be afraid. Like, I know I want to build something, I don't know what to build. Well, okay, a venture builder can give you an idea and run with it. And what they do is they give you resources, which is also helpful. You get some money, some people, so you can explore something. And then on my side, on the EIR side, it was like I needed a little bit of structure, a little bit of financial support. And I think that's also super helpful.

And also again, it's lonely at the beginning. And I think the other thing is that if you're over-optimizing on equity, that's another thing I think startup founders make—they over-optimize at the beginning. "Oh, I'm not going to get as much equity upfront or whatever." You know what? At the end of the day, most ideas fail. So great, I have a greater percentage of zero. And I think the idea with the venture builders, you get some ability to decide: is this a good idea or not? Is this really what I want to do? Does this make sense? Does this resonate with customers? And if it does, it gives you a bit of a head start there. And I think that's the advantage of if it's the right idea and it works. And by the way, I don't think people understand Snowflake came out of a venture builder model. Right? And so you can have hundred-billion-dollar successes out of a venture builder. People are like, "Oh, the biggest ones—" I don't think people realize Snowflake came out of an incubator and studio or whatever term you want to use. And so it doesn't mean—but I think to me, if you can build something that helps you build something successful, then that's what matters. What do you think is going to give you the best recipe for success? And again, it's not right for everybody. But if you want some support, if you feel like, "Hey, some nurturing, some structure, the ability—" and again, how do you help recruit that initial team? I think that's one of the things venture builders are really helpful for: helping pull that initial team together. Again, those first four or five people, it's hard to recruit people to an early-stage startup because you're getting paid about a third of what you should be. And you know, are you sitting in a coffee shop? Resources. But if you're in a larger group with a physical location and there's other people and startups and you can learn and reference and hear and get support, I think there's a lot of value there.

Amit: Right, right. And yeah, Snowflake was a great example. And you know, Kevin Ryan and his venture builder did MongoDB, Business Insider.

Todd: There you go. Another one.

Amit: Yeah, yeah. There's so many of them like Atomic has built out like four or five billion-dollar companies. Atomic, done, right? If you consider Atomic a venture builder, their entire portfolio is billion-dollar companies built up. Now, their model is to rip off American companies and build the European version of it. So be it. But, you know, yeah. No, this is great. Good to know.

Now back to Gumshoe. I wanted to ask you a little bit more about the product, if you can.

Todd: No, absolutely.

Amit: So basically, what we have seen in the AI visibility and discovery space is that there are basically two things, right? One is monitoring what is happening with your brand and with your products: how they are coming up in searches. And then the second thing is like fixing things and taking actions to get you to a better place. Can you explain in the context of Gumshoe: what are you doing—one of these, both of these—and what's the plan in the future?

Todd: Well, you know, our point of view on this is actually most of the value is on what we call the "now what" side, right? We view the monitoring as the goal there is to get a blueprint of now what to do. Like, now what? I actually need t-shirts that say "Now what?" Because when we talk to a marketer and you say, "Great, your competitors are crushing you in AI," and they're like, "Well, great, now what do I do?" Right? Universally that becomes the question. And if you look at the traditional SEO space—I was talking with an investor and his point of view is like, "Look, Todd, if you look at the SEO space, 98%—there's about $100 billion spent on SEO every year. About $2 billion goes to monitoring and tracking tools, and $98 billion is spent on SEO agencies and firms and services." So the market is telling you: 98% of the value is on the "now what" side, the "what to do" side. And so a lot of what we view and what we're building is on that actionability. So in terms of like our platform, we talk about yes, we help you figure out how to improve your website and what are the steps and things you need to do to go do that, and we help you do that. And then on the content side, one of the things that we do is we're analyzing those chats and using that as a roadmap of what you need to be talking about as a brand. And then on the outreach side, we're helping identify who are the people talking about this space and who you need to be mentioned by, and helping you craft those emails and manage the outbound of that. So we're very much in the belief that monitoring and analytics is only kind of the first step, the 2%, and the real value as a brand is on the "now what."

And even then, as we think about—what you know, I say at the beginning: we help brands understand what AI thinks about them. It's not so much AI visibility as much as it's you're trying to understand what does AI think about your brand and then what to do about it. And really what's fascinating, and that's kind of the lens today. But underneath that, you start reading the questions and answers. What it's really telling you is as a marketer: what does—not just AI think—what does the world think about you and your products? Because as you know with the personas I mentioned, we're helping you understand the personas of what AI thinks are your target customers and so forth. And what that does is you start to learn the messaging: well, what's resonating with these different audiences and so forth. And we have customers who are now taking the outputs of our platform—like our personas—and they're using that to help buy paid. Like, "Oh Todd, can I take these personas and use them for paid targeting?" I'm like, "You can do whatever you want. I'm just feeding information." We have other customers who are using it for product development. Like, "Oh, here's the strengths and weaknesses of my competitors. Great Todd, you're helping me identify where they're strong and I'm weak. Well, I can maybe redesign a product. It's not just talking about things. I might need to build something." Or if they're weak, maybe I can go do something.

A good example that we like to talk about is Lululemon. And it turns out they're doing really poorly with customers focused on sustainability. And what we're seeing is the upstart brands, the challenger brands targeting them, are really focused on that category—like sustainable fabrics and so forth. And it's kind of like Lululemon has a bit of the Nike/Hoka problem, right? Where the challenger brands—one's called Girlfriend Collective—in that with that persona and in that product segment, Girlfriend Collective now has visibility levels and recommendations and product adoption that are on par with some of the traditional Lululemon categories. And so we're seeing this for product development purposes. And so that's what I get—what I'm super excited about is as I see people leveraging our platform and using it for not just AI visibility but kind of marketing insights and marketing in general. That's where it gets really interesting and, I think, exciting. AI is really just a way to learn what the world thinks about something. And as a platform, our job is then to help marketers, "Okay, what do you do with this? What do you do about it?" We just got an inbound from people asking us, "Can I plug you into like a GenAI platform? I want to use your messaging to help me write generative videos." I'm like, "Sure." Like that's the type of stuff that gets really, I think, cool and fascinating. You start with AI visibility, but really it ends up with a platform to help marketers truly understand what the world thinks about their products and then what to do about it. That's how we think about the product and where we're going with it.

Amit: Yeah, very well put. And Todd, when did you start the company? And I know you have raised some money from PSL and some other investors as well. Can you just describe like where are you at and when did you start?

Todd: So we started a year ago, right? And I started as an EIR in late summer 2023. And then we launched the product in the end of February, beginning of March in public beta this year. So the product's only been—and we only went commercial about, we started charging about six or seven weeks ago. It's been fascinating to build. So we raised some pre-seed money, a syndicate which PSL was a participant in. We're just about to close our seed round right now. And so we have nine people on staff. One of the fascinating things is—you and I were talking, as background for this—what's really cool is all nine of us have been either a co-founder or a founding team member in a marketing tech startup. And I think that's been one of the keys here. We're really drawing on a wealth of experience of people who have sold into marketers for, probably combined, a total hundred years. And there's a lot of expertise there.

I think it's another thing about when you're building your founding team, right? And I think that's a better concept than just co-founders. Like who is that pre-seed core of 6, 7, 8, 9, 10 people? Your first marketing hire, your first sales hire, your first customer success hire, your first engineers. And I think, done right, you can really find people who have a lot of experience in your space. And I think that's super valuable, partly because if they have experience, then they've been through it before, right? They know the pain and suffering of what it's like to be at an early-stage startup. And for in this case, that's been one of the great things: our ability to pull this team together, all people who are used to the trials and tribulations of what it takes to build a marketing tech startup. So, as I said, we started a little over a year ago, we're now up to nine people, we're just about to close our seed round, and hopefully continue from there.

Amit: That's great, Todd. I wish you great luck and wish you all the best both in the company as well as personally. Thank you. And yeah, thank you so much for spending time with us. I learned a lot. I'm sure the audience learned a lot from this as well. Where can they find you? Usually, is it LinkedIn, Twitter, or some other place?

Todd: You can find me on LinkedIn and Twitter. On Twitter, it's @Swikipedia. It's my nickname. It's my last name—rhymes with "Wikipedia." And I'm good at trivia. And I won a trivia contest at a startup like 20 years ago. And apparently, there had been side bets about whether I'd answer the question because I had a bit of a reputation around my ability to do trivia. Well, it was always popular at bar nights, you know, at the trivia night at the bar. And I knew the weight of a gallon of water, and the reason I knew this was I wrestled in high school. And if you had to make weight, you used to spit, right, in the water fountain. You had to lose a pound or something, you know, and you knew how much you had to spit out to make that because you knew the weight of the water. And so, kind of an esoteric thing to know. So apparently there were bets. Like, my team bet on me and another team was against me. And so when I answered it correctly, my team jumped up. It was like cheers, you know, because they won a big—they won like a thousand bucks between them, right? And someone on my team said, he goes, "He's a Wikipedia." One of my team members said, "No, no, he's the Wikipedia." And that's where the nickname comes from.

So you can find me on LinkedIn. As I said, I connect with all my customers and users on LinkedIn. I'm active there. So yes. And you can, obviously, gum.ai. You can connect to us there as well. So Swikipedia, Gumshoe.ai, LinkedIn are all great places to find me.

Amit: I saw that your company was also called this by the same name. I saw it on LinkedIn.

Todd: Yeah. No, it's been a great nickname. And it's funny because, again, it allows people to know how to pronounce my name. And you know, I was always good at trivia. I don't know why, just always was. But it's kind of a fun thing to lean into.

Amit: Great. Okay. Thank you so much.