From Ad Tech Titans to AI Trailblazers: 7 Visionaries Building the Future of Advertising

Many ad tech veterans who previously dabbled in the startup life keep coming back to it. And with an AI bubble 17 times the size of the dot-com bubble, it looks like an ad tech-to-AI pipeline is beginning to form.

As AI sinks its teeth into every aspect of the tech world, from media buying to search, more and more ad execs are leaving their c-suite jobs, following their entrepreneurial instincts and launching their own AI startups.

Many of these founders worked in tech in the early days of the internet, and, they say, the dawn of AI is similarly scary, novel and unpredictable. But most of all, it’s exciting.

We spoke to seven ad tech vets about their forays into AI, and what set them on this path.

Jonathan Moffie, Streamr.ai

Jonathan Moffie’s history in ad tech is somewhat reminiscent of a Russian nesting doll. He started within larger companies and shifted to smaller and smaller ones, until he reached the center: his own startup.

Founded in 2024 by Moffie, Frank Turano (a former engineer at Shutterstock and Dotdash Merideth) and Burhan Hamid (former CTO of TIME), Streamr was established to make CTV advertising faster and more affordable for SMBs. In September, it was acquired by Magnite.

From 2022 to 2023, Moffie was head of product at streaming video distribution platform Wurl, before shifting to a small startup called Orka until mid-2024. Orka’s goal, Moffie said, was the same as Streamr’s: “to build the easiest way to advertise on TV.” Back then, he said, it was a lengthy process – and not a cheap one, either.

Streamr was born out of the belief that advertising on CTV can be done with any budget and in practically no time.

An advertiser just has to input its business name or website, said Moffie, and within a couple of minutes, the platform scrapes the site and online reviews and pulls relevant images and content to generate a commercial.

Within minutes, he said, the advertiser can hit “launch on TV” and, once approved by the streaming service, the ad goes live.

Moffie said his experience on smaller teams and working directly with other entrepreneurs is how he learned best practices for building his own startup.

“I made a concerted effort to join a startup” to learn from “people who had done it before,” he said.

Joe Luchs, Datalinx AI

Joe Luchs’ career has run the gamut of ad tech. He’s seen it all, from early-stage startups like Beeswax (which he helped to found) to what he called the “massive titans” like Amazon (where he served most recently as global head of AWS and Amazon Ads).

His latest project is Datalinx AI, which he described as an “AI data refinery” that takes raw, messy data and organizes it into “fuel” that companies can use, like clearly labeled and normalized datasets. Datalinx was co-founded by Luchs, Jeff Collins, Nicole Landis Ferragonio and Alek Liskov, who met through a mix of business school, AI-focused industry events and mutual friends.

A lot of companies have a whole lot of data but not in a form that they can easily access or garner insights from. It’s like if you bought a new car, said Luchs, but when you showed up to the gas station to fill the tank, you were told, “Hey, you need to drill into the ground, extract the fuel, refine the fuel, put it into the pump, then you can put it into your car and then you can go for a cool ride.”

Datalinx developed a team of specialized AI agents to complete tasks ranging from scanning a cloud database and processing datasets to identity resolution.

Each agent is powered by a series of LLMs, including OpenAI, Anthropic, AWS and Meta.

It’s a “very sexy solution,” he said, “to a very unsexy problem.”

Frans Vermeulen, Swivel

Swivel’s goal is to automate the entire ad operations process, from selling media to tracking campaigns to providing analytics and reporting, said President and Co-Founder Frans Vermeulen.

Vermeulen founded the platform with SpringServe co-founders Joe Hirsch, Matt Dearborn and Rich Lin. While “it’s still ad tech, and it’s still operations,” it’s “the most exciting thing [I’ve] ever worked on,” Vermeulen said.

Programmatic was designed to entirely automate digital media trading, and it did. But in doing so, it introduced new “operational burdens” that require humans to run and analyze reports – repetitive tasks, that could be solved with a “rules-based piece of software,” he said.

In addition to being repetitive, ad ops can also be frustratingly complex for marketers and publishers – the latter of which compose most of Swivel’s current client base.

Clients first have a consultation with Swivel to determine their goals. The company then selects a couple of use cases that it believes it can solve “right out the gate” to demonstrate Swivel’s ability to automate and replicate human behavior at scale, Vermeulen said.

This can look like anything from changing price floors or frequency capping to having varying strategies at different times of day, based on shifting demand patterns.

The platform makes “tens of thousands of micro-adjustments per day,” said Vermeulen, executing changes in “one fell swoop” that would take days for a human team.

“Humans are just naturally limited by the fact that we’re human,” he said.

Much to consider.

Michael Rubenstein, Firsthand

The “choose-your-own-adventure” format has come up in nearly every form of media, from video games to full-length novels and even that one episode of Black Mirror.

It only makes sense that ad tech would be next in line.

Firsthand, which markets itself as an AI-powered brand agent platform, is effectively a choose-your-own-adventure ad experience for consumers.

Brand agents serve as virtual brand reps who have knowledge of both a brand’s goals and individual consumers’ preferences, said Michael Rubenstein, co-founder and co-CEO of Firsthand. Those agents “guide consumers through a totally personalized, one-to-one journey.”

The AI agents process and understand content on a publisher’s site and tailor relevant brands’ ads to readers, with different product suggestions for viewers to choose from. For example, Firsthand’s agents might determine that a certain webpage is likely to reach consumers who have an interest in beauty products, so it would serve an ad for a beauty brand. But instead of standard contextual targeting, the ad would have multiple products to choose from, allowing the viewer to select the best fit depending on whether they’re familiar with the industry, curious about new products or shopping on a budget.

Once the viewer selects an option, they’re shown more pre-approved content from the brand with product information and recommendations on how to use it.

Rubenstein co-founded the platform with Jonathan Heller, with whom he worked at DoubleClick, and Wei Wei, Heller’s colleague at FreeWheel. (The circle of ad tech…)

“We love to build,” said Rubenstein, and “one of the biggest challenges in this new AI era is [determining how] brands survive and thrive.” They shouldn’t just be on the receiving end of AI, he added, but “users and beneficiaries” themselves.

Solving that problem with a brand agent platform, he said, is “the most interesting, exciting and important thing that we could be spending our time on.”

Rory Partalis, Classify

When Rory Partalis worked as Pandora’s ad product marketing manager in the early 2010s, he was fascinated by the platform’s Music Genome Project, which matches songs to an individual user’s tastes.

He later became VP of product at Partpic Inc., a startup that used computer vision and AI to help customers find missing parts of industrial products. Customers sent in photos of their products and the company, which was acquired by Amazon in 2016 and has since become defunct, determined the exact replacement needed.

“AI has kind of been sprinkled throughout the last decade or so of my career,” said Partalis.

So perhaps it was inevitable that his next step would be to found his own AI startup. Partalis launched Classify earlier this year, with co-founders Brendan Norman, a former Facebook exec and Nick Ross, the director of University of Chicago’s AI and machine learning data science clinic.

Classify’s “grand vision,” as Partalis described it, is to classify and contextualize all monetizable web pages across the internet. The tool scans over two billion web pages and pulls the content into its platform for categorization.

“We’re big on AI-first vibe targeting,” said Partalis. Advertisers can submit a single prompt and Classify will return relevant and high-performing content by understanding “what the customer is trying to target, who they’re trying to target, what their product is [and] what their goals are.”

Ultimately, Classify aims to help marketers advertise on relevant content and publishers package their inventory into curated deals.

The contextual targeting industry is built on “outdated taxonomies and categories and keywords,” said Partalis. More modern technologies, like AI, can look at a URL and scan a website in its entirety for a better understanding of the content.

Now, he said, is the “right point in time for privacy and AI to coincide and build something new.”

Joe Prusz, Medialive

Media buying has never been a simple process.

Medialive’s founding team set out to change that using – what else – AI.

A marketplace needs to serve both sides, said Joe Prusz, who recently joined the company as CEO and co-founder, and the goal behind Medialive was to “create a place for buyers, sellers and vendors to collaborate.”

(Prusz was formerly CRO of Magnite, and his co-founder Adrien Passerieux spent years working in sales and SaaS at Google. Prusz describes their collaboration as a “wonderful example of the buy side and sell side coming together to tackle a systemic industry problem that has plagued us all for far too long.”)

Medialive is like Slack or Whatsapp for media buying, Prusz said – a single room where everyone in the media buying chain, from the advertisers to the DSPs and SSPs to the publishers, can communicate directly.

At its core, it’s a straightforward system.

First, an advertiser uploads a brief into one of the digital rooms and invites other participants into the room who can converse and negotiate. Then, all parties can see the tech set up on both the buy side and the sell side. Medialive’s technology interfaces with APIs that pull in data from platforms including DV360, The Trade Desk and Magnite.

Medialive’s AI reads and summarizes the chat room in real time, said Prusz, and churns out a to-do list for the advertiser, from fixing geographic targeting errors to supplying missing inventory that could aid the campaign.

After spending over a decade at Rubicon Project (the artist currently known as Magnite), most recently as CRO, Prusz said he needed to build something from scratch. “That was really the entrepreneur in me,” he said. From there, he joined Hoppr, a small startup CTV platform, as CEO.

Prusz has seen the same problems across tech platforms for the past twenty years, he said: “broken workflows, the inability for humans to communicate effectively, mistakes on the front end that flow all the way through to the back end and create a wake of destruction for everyone.”

“We’ve lived these problems,” he said. And Medialive is trying to address them “all together in one solution.”

Kyle Csik, Adaly

It’s a pretty bold claim, but Kyle Csik believes his team has built “the Spotify of data.”

Instead of downloading data piecemeal and compiling it into a cohesive “playlist,” Adaly, co-founded by CEO Csik and CTO Aleksandar Sasha Grujicic, pulls in real-time data from all of a marketer’s sources. The company uses data from sources like Google Analytics, Statista and the Trade Desk to produces analytics and reports without any downloads or data infrastructure.

LLMs function as the brain of the system, Csik said: They can “think” and reason and connect disparate ideas. Adaly, on the other hand, he views as the nervous system, allowing the LLM to move in and out of ad tech and martech platforms. “Any model, any piece of data and any AI agent can be orchestrated in real time,” he said.

“The best use case,” Csik said, is asking Adaly to compare an advertiser’s own sales results to its competitors. Adaly will go into whatever platform an advertiser uses to track its own data, like Google analytics, and compare that against other sources like Circana data, churning out real-time comparisons and analyses.

This isn’t Csik’s first foray into automation. He’s worked at major holdcos and ad tech companies, including Publicis and GroupM, where the goal was always to develop new infrastructure and improve UI. Although, he thinks improving UI will become less of a focus as systems become more “destination-less” through APIs and natural language prompting.

Instead of requiring people to seek out a particular system, we should be bringing the system to them, Csik said. Adaly’s customers don’t need to go to the company’s website to use the tool, he said; they can call into the system directly from their phone, similar to how Siri works within Apple products.

Csik hopes that as more traditional UIs fade away in favor of natural language prompting, the quality of research and execution that Adaly provides will keep it afloat.

“I’ve built a career out of automating myself out of a job,” he said.