The Profile
Tadas Jucikas
He spent his twenties watching worms wriggle under a microscope. He spends his forties teaching machines what makes you buy.
He spent his twenties watching worms wriggle under a microscope. He spends his forties teaching machines what makes you buy.
Every month, Genus AI spits out more than 40 million product images and videos. They run on Meta, on TikTok, on Google, on Pinterest. They are the dynamic product ads that follow you around the internet after you glance at a pair of sneakers. The man who built the engine behind them does not come from advertising at all. Tadas Jucikas came from a laboratory bench in Cambridge, where the subjects could not click, scroll, or check out - they were millimeter-long worms.
Today Jucikas is founder and chief executive of Genus AI, a generative AI platform built for direct-to-consumer and e-commerce brands. The pitch is plain enough: feed it a product catalog, and it generates the images, the seed audiences, the copy, and the video needed to grow a brand across social channels. The interesting part is not the output. It is the worldview underneath it - that buying is a behavior, behavior follows patterns, and patterns can be measured, modeled, and predicted.
That is a thesis you can only hold with conviction if you have spent years reducing living things to numbers. Jucikas has. Before the ad creative and the funding rounds, there were peer-reviewed papers in Nature Methods and PNAS, and a strange, beautiful idea that an animal's entire repertoire of movement could be captured in four mathematical shapes.
Genus AI sells into a corner of marketing that is unglamorous but enormous: dynamic product ads. These are the catalog-driven units that pull a brand's products into a personalized ad, then test endless variations of image, copy and audience to find the combinations that convert. For a direct-to-consumer brand with thousands of SKUs, producing that creative by hand is impossible. Generating it with a model that has learned what performs is the whole game. Jucikas built a platform that does it at industrial volume - and frames the problem the way a scientist would, as prediction under uncertainty rather than art under deadline.
At the MRC Laboratory of Molecular Biology in Cambridge - one of the most Nobel-dense buildings on earth - Jucikas worked on the genetic and molecular mechanisms of behavior. The model organism was Caenorhabditis elegans, a roundworm beloved by biologists because it is simple enough to map completely and complex enough to be interesting.
His team built automated tracking microscopes and recorded thousands of short videos of worms moving. Then they did something quietly radical: they showed that the dizzying variety of how a worm wriggles can be compressed into four fundamental shapes - nicknamed "eigenworms." From those, they assembled a dictionary of repetitive behavioral motifs and 702 distinct measures, then linked them to the genes that shape them.
The work landed in PNAS in 2012 and Nature Methods in 2013. It is the kind of research that sounds esoteric until you realize what it taught him: that behavior, even messy biological behavior, has a hidden grammar. Find the grammar and you can read the sentence.
There is a second lesson buried in those projects, one that maps cleanly onto a startup. To study worms at scale, the lab could not rely on a human staring down a microscope. They built cheap automated tracking instruments and let software do the watching - thousands of videos, segmented and measured without a person in the loop. Automating perception, then turning what you see into structured data you can act on, is not a bad one-line description of Genus AI either. The worms changed; the method did not.
Earlier still came an MPhil in Computational Biology at Cambridge's Department of Applied Mathematics and Theoretical Physics, where he worked through machine learning, statistics and the modeling of biological systems. By the time he left academia, he had the rare combination of a hard quantitative training and an instinct for messy, real-world behavior - the two things that rarely live in the same person.
Reduce every wriggle, turn and coil a worm can make into a handful of building blocks. Recombine them and you can describe - and predict - almost any movement.
FROM 702 BEHAVIORAL MEASURES IN A WORM
TO BEHAVIORAL MODELS FOR SHOPPERS
After Cambridge, Jucikas spent years applying AI to human behavior, including projects for UK government departments - taking the same toolkit from the worm lab and pointing it at people.
He built and led data science teams working with major companies including Uber, EDF Energy and Tesco, pioneering real-world applications of machine learning across private and public organizations.
In 2017 he co-founded Genus AI with his brother Viktoras, who serves as CTO. They kept a serious engineering base in Lithuania while building the company toward the US market.
The path from a Cambridge bench to a marketing platform is not as wide a leap as it looks. The UK government work and the projects with companies like Uber, EDF Energy and Tesco were all variations on one question: given what a person has done, what will they do next, and what should you do about it? That is recommendation and prediction - the same machinery that decides which ad to show, to whom, and when. Genus AI is that thesis turned into a product, aimed squarely at the brands that live or die by customer acquisition cost.
What makes the company unusual is the order of operations. Most ad-tech starts with marketers and bolts on data science. Jucikas started with behavioral science and bolted on marketing. The platform's vocabulary - seed audiences, audience modeling, predictive content, lifecycle segmentation - reads like a lab notebook that wandered into a growth team. That is by design.
Genus AI raised in patient, unflashy increments rather than one headline mega-round. Early seed capital arrived after 2017; in 2023 the company closed a $6 million extension that brought total seed funding to roughly $11 million. The cap table reads like a tour of who pays attention to behavioral AI: Picus Capital, Transamerica Ventures, Aleph Group, and a roster of strategic angels including Oberlo co-founder Tomas Slimas - whose company was acquired by Shopify - and Treasure Data co-founder Kazuki Ohta.
The capital came with a stated direction. Jucikas has been explicit that the point is not to keep advanced generative tools locked inside enterprise contracts. The plan is to open the platform up, step by step, to smaller brands - the ones that have always been priced out of serious creative testing. In a field that loves to talk about moats, he keeps talking about access.
Running a generative AI company is not, apparently, enough to fill a calendar. Jucikas also serves as Founding Dean and Advisor at Turing College, an online institution in Vilnius that trains professionals in machine learning and data science.
It is a fitting second act for someone who keeps moving between worlds - academia and industry, biology and marketing, the US and Lithuania. The thread is teaching machines, and now people, how to learn from data.
It also says something about how he reads the moment. Plenty of founders treat the current AI wave as a land grab. Jucikas, who has watched the field for the better part of two decades, treats it more like a curriculum. Building the tools is half the job; making sure people can actually use them well is the other half. A dean and a CEO are not as different as they sound - both are betting that the scarce resource is not the technology but the judgment to apply it.
Put the pieces together and a coherent figure emerges. Not a marketer who discovered AI, and not a researcher who cashed out. Someone who took a single conviction - that behavior is legible if you measure it carefully enough - and carried it from a worm under a microscope, through government and the Fortune 500, into a company that now generates more ad creative in a month than most agencies make in a lifetime. The subjects keep changing. The question never does.