Jackson Engles, CEO of Minerva, is a realist. He knows that the data his AI startup is built on probably won’t be a competitive edge for more than a couple of years.
“Yes, sure, we have data that can help create better outcomes,” Engles told AdExchanger. “But is it a long-term, durable moat? Probably not.”
It’s an unusual thing to hear from the co-founder of a startup that recently raised $20 million to build AI agents for marketers powered by an identity graph of 270 million consumers.
So, if data isn’t the advantage, what is?
The answer, according to Engles, is to stay narrow. A specialized product built for marketing that starts with a curated set of data is more useful for brands than a general-purpose AI tool.
“What we’re trying to do, on some level, is to constrain the problem space,” he said. “Marketers shouldn’t just be using Claude and bringing in random data; they should be starting from a really opinionated view of the data that’s relevant to them.”
Marketing by accident
None of Minerva’s founders – Engles, Daniel Saedi and Mathew Joseph – have a formal marketing background. They met at Berkeley and started their careers in finance: Engles in banking at Lazard, Saedi in asset management at Bridgewater and Joseph as a quantitative researcher and data scientist at Citadel.
At Bridgewater and Citadel, Saedi and Joseph worked with web browsing and consumer spending data – known as “alternative data” in finance but standard issue in ad tech – to inform investment decisions.
Joseph helped build Citadel’s alt-data infrastructure, standardizing messy data sets and pulling out signals that the firm could trade on. They saw how valuable consumer data could be and how hard it is to make sense of it.
The three eventually left finance to start a company that would address that challenge. Getting into marketing wasn’t the original plan, Engles said, but it didn’t take long for it to become “the natural path.”
Minerva spent the first year and a half on the data itself, figuring out when people are in the market to buy and how life events impact their spending behavior. Then came the question of how brands could use these connections to drive commercial outcomes. The product grew out of that.
“The goal,” Engles said, “is to help [brands] get more out of their first-party data, better understand their customers and scale their marketing function with AI.”
Under the hood
Cool, cool. But plenty of startups make that pitch. Minerva says its approach is different, because it doesn’t solely rely on a brand’s own data.
The trouble is that first-party data only goes so far. Brands can’t see what customers do beyond their own channels, and even that data is hard to connect, because it sits in different systems.
Minerva says it addresses both problems. A brand plugs its CRM, data warehouse and other sources into Minerva, where an agent it built using OpenAI called the “agentic data engineer” maps it all out.
Then Minerva’s agents unify the data into a single view of each customer and supplement it with information from what Engles refers to as its “B2B2C identity graph,” which includes work email, personal email, phone number, address and online identifiers.
Ramp, a fin tech company and one of Minerva’s clients, used that link between work and personal identities to extend a direct mail campaign beyond work addresses, which boosted its marketing-qualified lead rate by around 80% in the first three months.
And for Wander, a luxury short-term rental company, Minerva built profiles of the brand’s best customers, used them to train its models and then sent lookalikes to Meta. ROAS was 3.4 times higher than with Meta’s broad targeting, according to Engles.
Minerva is also working on a second agent, an “agentic data scientist,” that would build predictive models from plain-language requests, like “Which customers are most likely to churn?” or “Who should I target with this offer?”
Like most AI marketing vendors, Minerva says its goal is to take the busywork off a marketer’s plate.
“Marketers should be spending way more time on the very human parts of their job,” Engles said. “They should be focused on telling the story of their brand, not the operational and optimization work that AI will get really good at over time.”
In my post-CDP era
Behind that pitch is a bet on where marketing software is headed, which is a world in which marketers talk to agents rather than log into dashboards. Consider it the post-CDP era.
Customer data platforms promised a single view of the customer, but left many marketers with just another system to manage. Minerva reimagined the idea as an “agentic CDP,” a middle layer that uses agents to join a brand’s data with Minerva’s so they can onboard quickly.
But Minerva doesn’t market the CDP aspect at all, Engles said, because marketers care about the end, not the means.
“Customers want the outcome of a CDP; they want their data to be more queryable by less technical users,” he said. “They want a single end-to-end marketing platform with its own data, not just a single capability.”
