Long before the AI hype kicked off in earnest, Eric Schwartz was already working with machine learning tools to develop optimized ad tech products.
At MiQ, he started playing around with automated custom buying algorithms, and then he helped launch AI bid optimization platform Scibids in North America in 2020. Scibids sold to DoubleVerify in 2023 for $125 million.
Now, he’s moved on from programmatic and into the world of performance marketing.
On Wednesday, Schwartz officially began a new role as CRO at AI-native customer acquisition startup Scowtt, which was founded last year by Eduardo Indacochea, a former top ad exec at Meta, Google and Microsoft.
Scowtt’s product offering can be broken down into two primary buckets: AI models and AI agents.
First, its AI models use first-party CRM data from brands – anything from order history to appointment times, call durations and message sentiment – to develop complex personas that can be loaded directly into walled garden tools, like Google Ads and Meta’s Advantage+.
The goal, Schwartz told AdExchanger, is not to replace search and social bidders, but to “feed those platforms with more information” to support effective bidding.
Second, Scowtt use AI agents as sales development representatives that can also use that CRM data to engage with consumers via SMS messages with suggestions of what to buy next, deals to entice lapsed users and other proactive calls to action.
AdExchanger spoke with Schwartz about why he joined Scowtt and his thoughts on how AI will impact the ad tech industry.
AdExchanger: For the past decade of your career, you’ve been focused on AI tools. Was that a conscious decision or did you fall into it?
ERIC SCHWARTZ: At MIQ, I wanted to give clients the opportunity to do things themselves. We had traders that were building decision trees manually and making optimizations, so I thought, how can we automate, save time and be cheaper, faster and smarter in the ecosystem?
That’s what brought me to Scibids, which developed AI and machine learning to automatically pull data from DSP ecosystems, learn from that data and push information as bidding optimizations back into the DSP.
Now when I talk to companies, unless there’s AI that’s driving the decisions and the learnings, it’s moot to me. I believe that just like how every company now has a website and chatbots, every company will have a personalized AI agent to engage with customers.
What is AI is really good at and what is it not so good just yet?
What AI does really well is to automate simple or mundane tasks, things like reporting and optimization in platforms. But if we allow it to do everything, we’re all screwed, right? I think there has to be a limit, and that limit has to be making lives easier for humans without completely replacing them.
Say you have 100 traders today. In the future, you could probably do the same thing with one quarter the number of people if you set it up correctly. But I still think there has to be a trader. There still has to be monitoring systems in place that humans oversee.
Meta’s been poaching a lot of AI leaders and making big claims about fully automated ad creation coming soon. Is that feasible and would it affect how Scowtt’s tools operate?
At Scibids, we saw that DSPs were getting better at optimizing towards their own sources of truth and updating their custom algorithms. What’s different here with Google and Meta is that we’re making their platforms better by enriching them with more data. We’re not in any way, shape or form competing with those solutions, which is very different from what I did at Scibids. So I’m not worried about that.
I do believe Google and Meta will continue to improve their AI for different features, but they appreciate what we’re doing. Google can go ahead and build a platform that takes CRM data and seamlessly puts it into a bidder so it has more information, but that will take two years for them to build with 100 people.
Why do I know that? If you look at our CEO Eduardo Indacochea’s background, he was a senior director at Google, a VP at Meta, and built a lot of the Microsoft tools. So we have a lot of intel and a lot of understanding.
You’re also an advisor to French AI startup Olyzon. Do you have any thoughts on how AI is being integrated into CTV buying as a larger trend?
What was attractive to me with Olyzon is that they’re able to enrich the bid request to target at an individual show level. If you have an AI agent that can help you pick out shows based on your RFP and make sure that they’re aligned, I think that’s really smart.
But there’s still a big gap in CTV when it comes to outcomes, like performance. Everything is PMP or direct buys. CTV is not new, but there’s still so much work that needs to be done in it, and I think AI will help people understand how to buy, measure and report on it better.
This interview has been lightly edited and condensed.
