Profound's James Cadwallader: brands are about to lose most of their consumer relationships to AI answer engines

We got you. Uh, you live? Can you hear us? Yeah, I can hear you. Yeah, there we go. Fantastic. We're live. What's going on? Cool. How you doing? We're good. Uh, can you uh start with just like a brief introduction of yourself, your company, what you're doing? Yeah, I'm James, co-founder, CEO here at Profound.

Uh, we're a New York City based startup. Um, yeah, we we're we're building a marketing platform that helps brands understand and control how they show up in AI responses. Mhm. Um, we launched in August, August 17th of 2024 and kind of like came sprinting out the gate.

Uh we we raised a $3 half million dollar seed round that was uh co-led by uh Keith Rabo Coastal Ventures with uh Saga uh uh sort of co-leading and yeah we've we've been growing pretty fast pretty quick. Um we have built this platform that's relied on by brands like MongoDB, Indeed, Mercury, Ripling, Ramp Ramp.

Um, yeah, we love ramp here. Um, yeah, talk about talk about like the genesis of the idea because I think what what you're doing and we'll get more into how it works, but what you're doing is now like blatantly ob like it's a blatantly obvious idea, right?

Because you all you have to do is kind of compare like as search became a big market and companies sort of rapidly realize like we need to understand and control how we're showing up in results.

you know, there's a lot of parallels to now, but I I thought your backstory of kind of like how you guys discovered uh the the opportunity in the first place uh at the beginning of last year was cool. Yeah.

I mean, it's it's it's you know, myself and my co-founder Dylan, we were at South Park Commons, um which is like a VC incubator accelerator. They have one in SF, one in New York. We were lingering around the one in New York um spending our time trying to figure out what business we wanted to work on.

And there was this kind of sort of ironic moment where as we were thinking about other businesses to work on, we kind of realized that we had just been using Perplexity all day, every day for weeks or months on end without actually going, you know, and doing traditional blue link search.

Um, and yeah, that again, it's one of those ones where, you know, it seems more obvious now, but this is back then, but even at the start of last year, chat GPT didn't have web retrieval. Uh, so it was no good at search perplexity. It felt like this streaming to CDs type moment.

And uh, we just went all in on that inflection point. We said, "Hey, you know, either Perplexity is just going to take everything here because this is so good and it this makes AI answers makes so much more sense than Blue Link search um or everyone else is going to jump on board and do the same thing.

" And probably closer to the latter is the way it unfolded. Um GPT introduced introduced web retrieval. Um I I mean Adobe did this huge study that they released like a couple of weeks ago. Uh it's based on like a trillion web visits on American retail.

Um and it's like I think 40% um of uh consumers now are are trying uh AI search for shopping uh for commercial purposes.

So, it's definitely We had somebody we had uh Aiden from OpenAI on yesterday who said that he just anytime he wants to buy a product, even if he's just buying hand towels, he just does like a deep research report and he's like it's worth just waiting five minutes to figure out like what's the best product based on my interest.

Uh I'm curious like how aware uh it makes sense that companies like RAMP are using these tools every single day and they start to realize like hey this is a product discovery mechanism. we should get on top of sort of understanding how ramp is is appearing in results.

But for how aware are legacy companies, you know, sort of like multinational public companies aware of how uh do they realize this is almost like a hair on fire problem yet?

Or or do you have to kind of explain to them like, "Hey, by the way, you know, 20,000 people are searching about your products every single day and like you have no grasp of of how you're appearing or or what that's causing people to do, etc. " They're all over it.

I I like reject this thing of the incumbents all had their lunch stolen in the early 2000s, and I think that everyone's like wiser now when, you know, they can sniff a platform shift. We work with some really big companies.

It's like I know everyone's got the, you know, the the founder yap, but we genuinely work with some of the biggest like Fortune 100 companies in the world. I just can't say their logos. Um because they just don't like that. Um but yeah, they're super clued up. They're super wise to what's going on.

Um they're maybe a bit slower in terms of the workflows because understanding is very different to like taking action and like trying to improve the results. But I mean, as an example, to give you one of our customers who I won't say who it is, but they're big company, 12,000 plus employees.

They did like a a big global PR push uh around this new product launch uh last I think a week or two ago and they saw a a 50% increase in their visibility on that like their visibility in answer responses.

W- without giving away the secret sauce, can you just give us some general guidelines for how a company should think about steering how they show up in LLM responses?

Because you could imagine everything from like do more SEO stuff, get on Reddit, make sure you're in the next crawl in the in the pre-training data set, or you could even go and say, "Hey, there's some there's some inaccurate information about us.

Let's just go directly to OpenAI and say, hey, can you trim this out of the data set? " Yeah.

So, I guess that there's a sort of wide aperture or sort of like philosophical response to this which is I think understanding that this is that the the failure mode here is to think this is like traditional search where I mean traditional search was essentially a marketplace system.

you had uh the website and the end user and the Google index just helped the end user you know come into a website maybe top a funnel and you know thumb around and the ecosystem eventually get pushed down the funnel into a conversion event AI answers there there's three parties at acting here where we have the we said the same website and the same end user but sat in the middle is the answer engine the answer engine changes everything because it's not connecting the too.

It the answer engine is essentially for all intents and purposes stealing the relationship from the end user. Mhm. Um or stealing the the the stealing the relationship with the end user from from the brand.

So I most brands don't aren't really groing this idea yet, but most brands are about to lose the majority of connection they have with their end consumer. Mhm. Because now when I'm researching a purchase, I go to chat GPT and if I need more information, I send a follow-up question to chat GBT.

There's no need to click out and eventually I mean, you know, if you read the the Strateery article that came out, the interview with Sam, I think this week or last week, you know, it's obvious that Open AI are going to go into shoppability. So, we're going to see like aic browsing, aentic shopping start to take place.

And you will you will transact within these answer engines. You will consume media within these answer engines. you won't need to go. We're going into a zeroclick future.

So, as a brand, the impetus, you know, going back back to that sort of philosophical point that the your impetus is how do I distribute information to these answer engines in real time accurately, the first step is like to just understand what the hell is going on.

So, you know, if I'm ramp and I want to show up more in accounts payable, cool. Let's like run a bunch of interrogations. Let's see how the answer engines respond when we send a huge variety of questions about accounts payable open-ended questions, not brand direct ones about ramp. Mhm. Let's see where they go.

Where which sources do they go does chat GPT go to or perplexity go to or co-pilot go to to get that information and then from there we can start building strategies. I mean and ramp you know they we just published a case study with those guys. So it's it's public information.

and they increased their um visibility like 7x in one of the indices that they were looking at by just looking at what was going on and what they could do about it. But then I guess John to give you a more direct response like you know tactical stuff llm. txt our our website tryprofound. com we get as our web our LLM.

txt now is getting as much traffic as our sitemap which is pretty interesting. We think for like very for rich extensive websites, you know, websites like I don't know Stripe as an example, we have tons of documentation. I think LLM's full.

txt will be very interesting because it's just essentially cramming a huge txt file of everything all onto one page because answer engines don't mind that.

Um I think going you know just sort of cycling through other ideas you know things that we're seeing work are lots of structured data um is very important like you know removal verosity you're creating content essentially for a retrieval agent to pick up not a human um images are useless uh answer engines don't care about images so like kind of ignore them to an extent.

uh metadata, you know, needs to be very thoughtful. Like you can essentially you can you can just spoil the whole article with your metadata and that will often be like pretty effective. So I mean I could like I could go through I could cycle through. Yeah. What uh I have a couple kind of questions.

You're sort of building in a what's a new category? Do you call it a AEO like answer engine optimization? Do you have you coined it yet? Obviously you want to own it. Yeah, it I think you have to let the kind of market decide what it's going to be called. AEO, AIO, LMO, um, GEO gets thrown around.

Generative engine optimization gets thrown around a lot. Everyone, considering we're all like techno optimists and, you know, Silicon Valley is meant to be full of technoptimists. It's the the desire to use heristic is quite interesting to me. I've seen like so we as human beings, we love to use huristic.

we just want to say this is SEO for AI, but there's all kinds of like new surface areas that just didn't exist before. Um, yeah. Uh, can you talk about do you have any type of vision of what paid ads in LLMs will look like?

I had gotten multiple pitches over the last year of people that were building these sort of like ad networks to fit within uh different LLMs.

And my general read is like I sort of believed that ad, you know, paid ads would be heavily integrated into LLMs, but I sort of felt like it might I wasn't necessarily convinced that in a world where there's, you know, an infinite number of different answer engines that everybody has a different favorite.

Maybe the sort of global network makes sense. But then if OpenAI is getting like, you know, long-term or Grock or whoever's like the sort of power law winner, if if they're getting 80% of the traffic, like maybe they end up with more of like a meta style ad network where they just build it, you know, from scratch.

Um, but I'm curious what you think. I mean, I definitely uh I kind of think well I I I believe there will be a more fragmented uh marketplace this time around. I think there's just more everyone knows what's at stake here.

uh Google search has been the biggest monopoly in the history of the internet and I think there's more capital more energy going into trying to ensure that doesn't repeat itself this time round. Um but answering the question I I think it's actually two things.

I think it's generative advertising in a conversational user experience will be that the convergence of those two things will be one of the most powerful unlocks in the history of marketing. Language models are really good at language.

Um and being able to synthesize net new ads on the fly that are perfectly tailored to that exact moment in the exact conversation in the exact way. Yeah. that is most compelling I think is about to melt marketers brains on and on the efficacy of like how how powerful that advertising medium is going to be.

I I mean I I'm I we want it to happen. We think it's very interesting and we think it will speak perfectly. It will be a very obvious next move for our business. Um I don't know. We're not holding our breath right now.

I don't think I you know I don't I'm not sure if it if if it will unfold like that especially with open AI you know they have that corporate the the corporate structuring I think is a bit of a hindrance just to kind of the the values of the business I can't I don't know your speculation is as good as mine but yeah yeah well very cool uh I'm sure uh I'm sure you're going to be raising a lot in the next few years so uh when when you're when you're ready to announce anything come back on and break it down send me keep sending me different learning ings and stuff in the meantime.

I think we we just have a ton of founders uh in our audience that probably aren't taking LLM optimization seriously enough now. So, uh go uh DM James or or sign up for a demo or whatever. And uh it's great having you on. Thanks for having me, guys. Yeah, congrats on everything. It's cool. Yeah, thanks for stopping by.

Later. We got we got Quaid in the in the waiting room yet. We got Quaid coming in to the Temple of Technology breaking down watches and wonders which of course is it seems like the biggest watch event of the year. Uh every major watch brand announced new watches and we're going to get the update from Quaid.

I'm very excited for this. We discussed the Vashron Complications watch that launched but I want to hear about PC. I