She left an insurance underwriting desk because she was sure a machine could score credit better than a committee. She was right. Then she did it again - for advertising.
SOOWON "SOPHIE" EOM // FINANCE → AI → MARTECH
Right now Sophie Eom runs Adriel, an AI marketing-intelligence platform that pulls scattered ad data from a dozen channels into one live picture. The pitch is unglamorous and exactly right: stop screenshotting dashboards at midnight.
Marketers live in a hall of mirrors. Spend goes out across Google, Meta, TikTok, YouTube, and a long tail of channels, and the numbers come back in different shapes, on different clocks, in different tabs. Adriel's job is to unify the pipes, surface what matters with AI, and let a person make a decision before the budget is already gone. Today the company counts more than 250 enterprise and agency customers across over 20 countries, with around 60 people behind it and roughly $5.9M in revenue reported for 2024.
What makes Eom unusual is not that she founded a martech company. It is that she arrived from finance - HEC Paris, Oliver Wyman, AXA - and treats advertising the way a risk analyst treats a loan book. Measure. Predict. Optimize. Repeat. The vocabulary is marketing; the instincts are underwriting.
AI-driven marketing intelligence. Real-time cross-channel reporting, automated alerts, and creative insight - headquartered in Austin, Texas.
Her first startup: machine-learning credit scoring for big banks. The work that put her on the Forbes list.
There is a tidy logic to Eom's career if you read it backwards. Underwriting is the art of turning incomplete information into a confident decision: you gather the signals, weight them, and act before certainty arrives, because certainty never does. Advertising has the same physics. A campaign is a bet placed on partial data, judged on a delay, and corrected - if you are lucky - before the money is spent. Most marketers are flying that plane with the instruments scattered across a dozen browser tabs.
Adriel's answer is to collect the instruments in one cockpit. The platform plugs into the channels a brand already runs - search, social, video, programmatic - normalizes the numbers so they finally speak the same language, and then layers AI on top to flag what a human should look at first. Automated reporting replaces the Monday-morning ritual of stitching screenshots together. Alerts catch the spend that is leaking before the month closes. White-label dashboards let agencies hand clients something that looks like clarity instead of a spreadsheet apology.
It is a deceptively boring problem, which is exactly why it is a good business. Eom has said the first step for most advertisers is the one they skip: actually defining the customer they are chasing. You cannot optimize toward a target you have never described. The technology is downstream of that discipline - the dashboard only matters once you know what the numbers on it are supposed to mean.
Spend scatters across channels that each report differently, on different clocks. The truth arrives late and in pieces.
Unified pipelines, AI-surfaced insight, automated alerts and reports - so the decision happens while it still counts.
Don't cut your marketing budget just like that.SOPHIE EOM, ON SURVIVING A DOWNTURN
After HEC Paris, Eom did the sensible thing first. She consulted at Oliver Wyman, then moved to the insurer AXA. Inside that world she noticed a problem hiding in plain sight: credit decisions leaned on coarse rules when the data could support something far sharper. Machine learning, she figured, could give financial firms fairer, more accurate scores than a room of underwriters working from a rubric.
So in 2014 she left to build it. Solidware turned that hunch into a product - predictive analytics sold into large financial institutions. By 2017 the bet had earned her a spot on the Forbes Asia 30 Under 30 list, the kind of recognition that tends to arrive once a contrarian idea has started printing results.
The pivot to Adriel in 2018, co-founded with Olivier Duchenne, was less of a leap than it looks. Credit scoring and ad optimization are the same shape of problem: messy inputs, high stakes, decisions made too late on data that was never clean. She swapped the loan book for the media plan and kept the method.
Solidware also taught her something about selling to skeptics. Large financial institutions do not adopt a startup's algorithm on charm; they need the model to be legible, auditable, and right often enough to matter. That habit of making the machine show its reasoning carried straight into Adriel, where the customers are agencies and enterprise teams who have been burned by black-box tools before. The product earns trust the same way the credit model did - by being explainable, not just clever.
Geography is part of the story too. Eom built her early reputation inside the Korean startup ecosystem, where she has been cited among the women leaders worth watching, and she has carried that across continents. Adriel today runs from Austin, Texas, with a team of roughly 60 and a customer roster that spans North America, Asia, and Europe. The company she leads is, in a sense, a working argument that marketing problems are not regional - bad data looks the same in every language.
Adriel's capital came in stages, the way conviction usually does - a little proof, then more money, then more proof. Total raised lands around $20M, with a Series B in March 2022.
Between shipping product, Eom kept turning up in rooms where the agenda was the future of everything.
Member of the UN Secretary-General's High-Level Panel on Digital Cooperation - seated alongside the likes of Melinda Gates and Jack Ma.
Served on the Presidential Committee on the 4th Industrial Revolution, 2021-22, advising on the policy edges of AI and tech.
Listed in 2017 for Solidware - recognition that the credit-scoring bet had started to land.
"Many studies have consistently shown that brands that continue marketing in a recession are the most likely to survive."
"AdOps is basically a system and culture in an environment that can support the whole process. It helps collaboration between different teams."
"Surprisingly, there are so many clients and advertisers who cannot concretely define their dream customer. But building out a detailed customer persona is the first step."
The user journey, she points out, is now hopelessly tangled - YouTube at night, Instagram for photos, a different app for chat - which is exactly why single-channel strategy is dead.
"AI in big banks" - Sophie Eom on the Solidware era and bringing machine learning to financial institutions.
Eom frames Adriel's ambition in bigger terms than a reporting tool. The goal she describes is to set a new standard for marketing intelligence - to make complex data pipelines simple, to surface insight with AI instead of armies of analysts, and to let marketers move at the speed their channels actually run.
That is a tall order in a category crowded with dashboards, and it is the kind of claim that only the work can settle. But the through-line of her career suggests she is comfortable being early to an argument. She bet that machines could underwrite credit before banks believed it. She bet that fragmented advertising deserved a single source of truth before most teams felt the pain acutely enough to pay for it. Both times the bet was the same: that judgment improves when the data finally stops lying by omission.
There is also a quieter thesis underneath the product. Eom keeps returning to the idea that the user journey has splintered - people drift between YouTube and Instagram and chat apps without warning - and that no single channel can describe a customer anymore. If that is true, then the future of marketing is not better creative on one platform; it is better synthesis across all of them. Adriel is her wager on that future, and on the unglamorous infrastructure that makes it usable. For a founder who started by questioning how banks scored a loan, it is a familiar move: find the place where everyone accepts a messy guess, and replace it with something you can measure.