Few brands have the time — and pocketbooks — that helped forever link Reese’s Pieces to the most famous extraterrestrial in movie history. But the AI-powered ad tech company Rembrand found a shortcut: a way to place a soda can, conditioner bottle, or whitening pen into digital video with AI, even nailing the scale, sizing, and depth so well that viewers remember the products after a show has ended.
Even more critically? Rembrand CEO (and ad tech pioneer) Omar Tawakol says that viewers are less likely to speed up commercials they see later when those products appear, raising “the efficacy of the rest of your ads,” he says.
“It’s a very compelling consumer vision,” says Tawakol. “In the long run, consumers win.”
You can hear more from this conversation between Tawakol and Signal 360’s John Battelle in the video below or read our lightly edited transcript.
TRANSCRIPT
John Battelle
I’m pleased to be joined by Omar Tawakol, who is the CEO and founder of Rembrand. You may know Omar, and you certainly know his companies, including BlueKai, where he was the CEO and founder, a central player in the programmatic advertising space acquired by Oracle and became central to Oracle’s Data Cloud. After that, Omar started an AI-voice company. This was before the large language models. So Omar was early in the AI space. And that company, called Voicea, which was acquired by Cisco. He’s now starting another AI company, Rembrand, and that’s what we’re here to talk about. So welcome to Signal, Omar.
Omar Tawakol
Thanks for having me. Glad to be here.
So am I. Thanks for joining. Your career in tech and advertising is pretty extensive. You’re very well known in that space, and it’s culminated in a company, Rembrand, that when you told me about it, initially, I was, “Wait, that just doesn’t sound big enough for you.” You do these very big, very important things. But over the past year or so, I’ve realized, it’s a big deal. So can you tell me, what is it that Rembrand does?
Very kind introduction, John. I think if you think about what we do, we were part of watching programmatic advertising go from 0 to $250M, and it was just an amazing ride. But we’ve become so efficient at targeting every ad to the right person, to the right context, with the right data. But what has happened over the past two decades? If you walk into the room and talk to most executives who allocate ad budget, you ask them what they do when they go home, they literally go home and pay money to avoid ads. They watch Netflix, they watch all these other shows. They pay YouTube for premium. They’re not alone. Consumers are figuring out how to either skip or pay to avoid. So why bet against consumers? What we at Rembrand figured out was, how can we use AI to integrate a brand into the content people love without stopping the show or annoying the audience. It’s a very compelling consumer vision. In the long run, consumers win.
We didn’t want to reinvent the wheel. Product placement has existed for decades. I mean, think Reese’s Pieces and “ET” to date ourselves, but you know, that was super impactful integration. For decades, there’s been this great product placement. The problem is, if you want to get into a movie, you’ve got to wait two years in advance of the movie to negotiate to get into the script. It’s just such a manual process. With AI now we can take any video in a catalog, in a movie catalog, there’s already filmed or any new TikTok, Instagram, YouTube or LinkedIn post, and we can post production, put the right brand in there so that the consumer can continue to enjoy the content, but the brand can get attention. That’s what we do, use AI to make product placement look like a media buy.
I want to get into this a bit, because once you open that Pandora’s box, you realize that you have to build, or at least integrate with, a lot of things that may not have been there before. If Reese’s cuts a deal with, Spielberg’s production company to get Reese’s Pieces into the movie, that’s great, and they’ll probably see a bunch of sales lift because it was a hit movie that they didn’t know it would be two years beforehand. But when you’re talking about the entire catalog of video, both that is in a library, or that might be created on the fly, you’re talking about a whole media ecosystem that has one way of measuring impression and efficacy and return on spend and all of that. How do you think about that when you’re talking about something that isn’t a traditional ad unit, so it doesn’t flow necessarily on the same rails of measurement as a display ad or a 15-second spot?
There’s a lot of magic that has to be figured out exactly there. You’re putting your finger on the pulse there. First of all, what corner of advertising are we really after initially. It’s essentially brand lift. What we’ve done is done is Kantar study, after Kantar study, literally, a ton of them. Every single one of them had a very large lift in unaided brand awareness. Then you get some lift and purchase intent and aided brand awareness. The beauty of unaided brand awareness is you’re asking them what brand you saw, and you don’t give them a drop down box so that they have to remember it. So we do really well there.
The other area that we do well is we’ve measured if you retarget a user who saw a Bubbly can or one of these products sitting in a 30-minute episode, if you later target them with their standard ad on YouTube, what you find is the skip rate of those ads plummets because of the priming that’s happening for that user having seen that product in front of them. They just take longer to skip, or they don’t skip. It raises the efficacy of the rest of your ads. That’s really where we’ve been going after large brand advertisers.
The next phase of this is really to go a little bit further in measurement, offline sales measurement and connection to shopability. That’s the one of the extra areas of gold I’m really after. Because when you make it much easier to buy through a shoppable link, so you’re starting to see variance of that between YouTube and TikTok, that’s really interesting.
I would imagine that the actual interface of this, as a consumer of that 30-minute episode, for example, and there’s a product in there. Do you think behaviors might change to the point where people want to interact directly on screen with the product, and that placement might become or is it more in the Chrome around?
I think it’s more in the Chrome around. The exception is with certain types of influencer videos where you’re watching it and you’re like, I just like John Battelle glasses. Show me where I can buy that. There’s some of that in the influencers, less so in like movies and TV. But there are some techniques you can use with movies and TV, where you somehow take some of the imagery from the show and do something with it after in an ad. I’d say, in the next year or two, we’re going to see some innovation on the connection. Right now, we’re just crushing it on brand lift. But this next phase of showing either offline sales lift or direct engagement as you’re saying, in the Chrome around it, that’s gonna be pretty interesting.
Part of that has to do with scale, right? When did Rembrand first go live?
Rembrand first went live in the spring of 2023, the very beginning of 2023. We started with brand advertisers, later in the year. So I’d say, we’re about, 16 months or so into large-brand advertising campaigns.
There’s clearly a place that I’ve heard a lot, and I’m sure you have too, of concern from traditional brand marketers that they can’t fill the top of the funnel. They can’t create awareness and preference. The bottom of the funnel is almost over filled with solutions, but the top of the funnel isn’t, and this is clearly, as you’ve said, a place that addresses that. I’m curious, what about scale? Because when brand marketers think about creating lift and awareness, they think about Super Bowl ads.They think about sports.They think about big audiences. How are you doing on the scaling journey?
The scale has been surprisingly good, because we did something really counterintuitive. Everybody who approached virtual product placement before us started out saying, “Well, of course, there’s the Reese’s Pieces example. Let me see if I can get into a really famous movie.” Startups walking up to the biggest movie studios saying, “Let me take your movie, alter it, and sell it.” It’s hard proposition. So usually what they’ll answer them as saying, “Yeah, cool idea. Here’s a small little movie or show that played in this small, little province in some other country. Go ahead and sell that”. So you have all these scale challenges, and the reality doesn’t meet kind of expectation.The other problem is, before using the AI that we built, people would actually have to use manual VFX to, frame-by-frame manipulate. And there’s 32 frames a second. So this is a lot of work to do there.
The counter intuitive thing we did is we said, postpone movies and TV. We’re now in movies TV, but the first year and a half, we didn’t touch that. We just went to influencers. The amount of incredible scale you get when you go to these large influencers with 10s of millions of followers. There are some episodes that they will put out that if we put a product in it at the CPM rate that we pay them, they would eat up the entire campaign budget. We had this one influencer, just a few million in reach, he had an episode, he put out, he made $50,000. All he did was, he finished his episode, it was an episode about dating, and came to us, pulled a virtual product off the shelf, put it in, in the speed of compute, he uploaded, and bam, he made $50,000. There’s a lot of reach there. What we did is we started getting that high reach and reasonable number of videos to train our algorithms. Now that we’re entering into the sexier domain of movies and TV, now we’ve got the automation and the AI algorithms to take it on, as opposed to manual techniques.
I imagine that this shift. I mean, it’s not really a shift the addition of traditional movies and TV, and in particular, streaming, as the vector for Rembrandt offerings has got to be making you very busy. Because it’s a little different working with influencers and working with Amazon Prime, and you know, Apple TV, and folks like that. What are you finding in that market, because I know you’re a great observer of markets, is different and might be a challenge in terms of adoption of innovation.
When you list all the top streamers, as you said, Apple TV, Amazon Prime, Netflix, Disney, those are the streamers. I’m not saying those are my customers. What do they care about? They care about the quality of their experience, the quality of their shows, and then the ability to monetize it themselves. Those are not the kind of players where you can walk in as a startup and say, “Hey, I’m a I’m a network of VPP, let me sell your stuff.” They’re going to say, “Well, there’s the door.” What they want would be technology companies that let them control. It’s very important to have technology enablement as your entrance into this market, and not the easy thing, which has become an ad network, and hope the biggest guys come to you.
If you think about the development of the ecosystem that we saw and the generation before. The reason kind of DoubleClick and Google had some strategic options, is DoubleClick had an ad network at one point. Then I had DFP (DoubleClick for Publishers), which is publisher-side, then DFA (DoubleClick for Advertisers( that was advertiser-side, and then they created an Ad Exchange. That kind of strategic flexibility is important if you’re entering a market. An ad network-only model is easy to build, but very quickly it gets commoditized by technology players.
Are you coming in almost as a technology enabling layer? They’re going to take their sales force, their business development teams, their partnerships, leverage your offering, as opposed to you just going to them and saying, “We’re going to bring the clients to you.”
Correct. We entered the marker in the first16 months with a network model, meaning the biggest brands buy product placement on social. That’s our current model. But starting a few months ago, we started going back to all the players, saying, “The longer we’re here, we just want to do the software.” We’re going back and now giving software to a bunch of players now that we’re in CTV, and it’s very early days, very early. We just entered into CTV Q4 of last year. That’s where we’re more aggressively going with this software model to enable the players and to make it clear we’re not here to compete with our customers.
If we could pull back, I want to just spend a minute or two on on AI, generally, you’re obviously deep in that space. You understand better than the average person our audience. There are a lot of people in marketing, a lot of people who manage brands, what should they be paying attention to right now? There’s so much noise. What are the things that they really should be focusing their attention on?
I basically start out by saying, awhile back, almost like 18 months ago, Terry Quad just said something really interesting. He said, the whole investment level in the ad tech ecosystem, has been about efficiency, squeezing bits of efficiency out of better bidding, better data, better targeting, header bidding, all this kind of stuff. Billions went into support companies that were doing this, and billions remain. If you think of that as an iceberg, efficiency is the top of the iceberg above the water. His visualization, that was really cool, was to say there’s actually, instead of little 10% improvements here at the tip of the iceberg, there’s, hundreds of percents of improvement, if you change the actual creative itself, the thing that’s engaging the mind. Typically, those two manuals the realm of human creativity, now with AI, it’s almost we can bring that water level down and automate the act of creativity, not completely, but that’s where there’s like a Cambrian explosion of activity, of where you can apply AI to do very interesting things in marketing. That’s a high level. But at a more detailed level, if you’re a marketer, what are the practical things that are being attacked or kind of being looked at? I’d say obviously, like custom versions of ChatGPT that allow you to do things that are more in your brand voice, like email, copy, ad copy, blog content. There are companies like Jasper, in addition to ChatGPT write it out AI. In the visual space, obviously Rembrand with VPP. If you look at Memorable which just got bought, they were helping you determine which creatives were good, which parts of the ad are going to get seen by modeling eye tracking. They did some really cool things. There you have some VFX automation that’s happening with runway. I think those are practical areas you can start to look at in both marketing and creative.
I don’t want to let you go without asking you about there was a very significant flurry of attention given to a new AI model that came out of China, DeepSeek. I’m curious if you think it’s as revolutionary as the headlines seem to make it out to be.
I do, because I was kind of confused. It was January 3, when the first information came out of that, and I started using it, and I’m like, “Why aren’t people completely going nuts about this?” I didn’t understand. Because when they talked about a training run for $5 million I thought that was super significant. But there was, in my opinion, a lot of good to it. First of all, the fact that it’s open source, and it gained from open source is itself really good. If you look at what Yann LeCun was saying, this is a victory for open source.
Of AI at Meta.
Yann LeCunn is one of the three fathers of AI, along with Hinton and others. He was talking about this being a victory for open source. I agree, and obviously he has a horse in the race. But he wasn’t poo pooing them. He was saying, ‘This is open source, and it’s a victory for it.” We shouldn’t be just poo pooing something because it’s coming from another culture or another, another kind of competitive economy. I agree that maybe there’s some exaggeration to the cost side of what they did. There’s some work that needs to be done to validate what it really means when they say 5 million, what’s included, what’s not. Barring that, they definitely were very creative and necessity is the mother invention, because they had constraints. I think those constraints are really good.
The second thing that’s really good is that every empire and every company that wants to become an empire benefits from scaring the hell out of everybody into believing that resistance is futile. Do not compete with me. Too many people have been laying over, playing dead, thinking, “Okay, I can’t compete.” Here these guys came in and showed even at the foundation level, you can compete with creativity, right? I think this is a great thing. The other line of attack people are using against DeepSeek, which I believe is really questionable, is they’re talking about, “Look how the Chinese are going to use our data.” Do you remember when we were caught spying on Angela Merkel’s phone in 2013. Come on, guys. I understand that the Chinese government has policies that means they can do whatever they want with our data. But let’s not be all high and mighty here.
DeepSeek points into a direction where thousands are already swarming to recreate it and and reset and level set what’s possible with models that have all sorts of different attributes and parameters and data sharing, frameworks. To me that sounds that seems very exciting, because then you can really level up to what’s the product here, and what’s the product that gets built based on the economy’s downstream and upstream that were impossible prior, or, as you pointed out, presumes to be only products that would be offered by the big companies. I like to see the ecosystem develop, and I like to see companies like Rembrand take advantage of it. Omar Tawakol thank you so much for being part of this Signal Conversation, and we’ll look forward to checking in with you as you develop into yet another powerful company in the marketing and advertising technology stack.
Thanks, John. Great chatting.
