| | | Plenty of founders say they built something to scratch their own itch. Selina Tirtajana spent a decade inside medical device companies, found nothing that worked for her own teeth grinding, and built the dataset, the sensors and the company herself. | That mix of stubbornness, scientific rigour and a very personal reason to get it right is why she’s in our 2026 class of the Most Interesting Women in AI. | -MV | What’s in store: | | Read Time: 6 minutes |
| |
|
| | | | | WOMEN IN AI 2026 | The Woman Who Built the Thing She Could Not Find | | If you have ever woken up with a jaw that aches for no reason you can identify, Selina Tirtajana has been thinking about you. | She has also been thinking about herself, because she grinds her teeth and has spent over a decade working on medical devices without finding a single solution she could actually use. | This bothered her enough that she took matters into her own hands. | Selina grew up in Indonesia and Singapore in a family where her mother and sister are physicians. The kind of household where medicine is not a career option so much as an expectation. | She was the rebel in the family. So she went to Wesleyan to study liberal arts then Yale to study public health, spent a decade inside large medical device companies like Covidien and Medtronic learning exactly how slowly things can move, and then founded her own. In a way, she never left the family business. She just took a longer route there. | SØVN — like the Danish word for sleep — is a set of AI-powered earbuds that detect teeth grinding during sleep and stop it using in-ear magnetosensing and biofeedback: a gentle signal that prompts the jaw to relax without waking the wearer. | It's a very selfish invention. |
|
| Selina built it with Gabriel Saenz, her partner, out of Utrecht in the Netherlands. The device tracks jaw movements continuously through the night. When it detects grinding, the biofeedback signal is subtle enough to reach the nervous system without pulling you out of sleep. It sounds like science fiction, but it's just science, really. | You wake up without the ache. It's "a very selfish invention," she says. | The thing about teeth grinding that most people don't know: it's not actually about your teeth. | "It's not a teeth issue whatsoever," Selina says. "It's not a muscle issue. It's our autonomic nervous system responding to perceived danger while we sleep." | It happens during unstable sleep, in the transitions between stages, when the body is doing something the conscious mind has no access to. Which is to say your jaw is clenching automatically and you're not even awake to stop it. Night guards protect the teeth from the damage, but they don't stop the grinding. And Selina, well, she wanted to know how to stop the grind. | It's not a teeth issue whatsoever. It's not a muscle issue. It's our autonomic nervous system responding to perceived danger while we sleep. |
|
| Before she could train an AI to detect it, she had to build a dataset large enough to train on. The problem was that no such data existed. The largest available data on teeth grinding was based on fewer than fifty people. (A very small number for a condition affecting one in ten adults, by the way.) | Selina spent three years developing the sensor technology before they could start collecting data at scale. The research is ongoing, in collaboration with technical and research universities in the Netherlands. A patent was granted in 2023, the same year they won the CES Innovation Award for Wearables. | The company was originally called Jawsaver, but people kept calling it Jawbreaker. This happened enough that Selina renamed the company rather than keep correcting people. SØVN cannot be confused with hard candy. She considers this progress. | Her vision is for SØVN to become a global sleep health brand, moving beyond teeth grinding into the wider space of keeping the body in the most restorative state possible during sleep. The international market is wide open, she says, with the confidence of someone who has already done the research. | She built the dataset when there was no dataset, and she definitely is not waiting for permission to do the rest. | If you want to stop grinding your teeth, Selina's here to help. Find her at getsovn.com. | Get 22 AI agents you can deploy in 5 minutes below! | |
| |
|
| | | | | POLL OF THE WEEK | What’s the biggest AI disappointment so far? | | Full results will be at the bottom of tomorrow’s newsletter! |
| |
|
| | | | | THIS WEEK IN AI | Blue waves, billion dollar IPOs, and Jensen Huang’s very bad month | | Democrats swept the midterms and Nasdaq lost $2 trillion before the week was out. Anthropic is going public at $1.4 trillion with a $104 billion net loss. And Jensen Huang resigned as Nvidia CEO after 33 years following a spy scandal and an affair. | Completely normal November energy. | Here’s what mattered: | Democrats won control of Congress in a blue wave midterm sweep, triggering a Nasdaq sell-off of more than 5% and wiping $2 trillion in market value as investors reacted to plans for independent AI testing, liability standards and data centre guardrails. World Gymnastics is in talks to bring Fujitsu’s AI judging system to the 2028 Los Angeles Olympics, using four cameras per event to build 3D routines and measure joint angles to the degree, with full results delivered within 15 seconds of landing. Nvidia launched an investigation after reports claimed CEO Jensen Huang was targeted by Chinese intelligence through a former Megaspeed executive, with US lawmakers requesting a private national security briefing on November 18th. Adobe’s Creators’ Toolkit Report found 78% of AI-using creators say it helped their business grow faster, while 47% abandoned an AI tool mid-project in the last three months because getting the results they wanted took too much effort. Anthropic plans to sell 825 million shares priced between $70 and $78, potentially raising $64.35 billion at a $1.4 trillion valuation, with shares expected to begin trading on Nasdaq on November 19th under the ticker ANT, backed by $26 billion from seven major tech companies. Adobe’s MAX 2026 conference expanded its Creative Agent across Photoshop, Premiere and Firefly, letting users describe tasks in plain English and have AI execute multiple steps at once, alongside new marketing AI agents and broader Google and Microsoft integrations. Google DeepMind unveiled AlphaGenome 2, trained on over 100 million cell profiles across 68 tissues, identifying nearly twice as many genetic variants affecting gene expression as the previous leading model, with free access for non-commercial research. OpenAI acquired Odessia, an AI travel startup that helped users plan and book trips before shutting down, with founder Francis Davidson joining the company alongside existing partnerships with Booking.com, Expedia and KAYAK. Researchers at Physical Superintelligence used an unreleased Anthropic model to solve the Aaronson-Ambainis conjecture in just three days, an 18-year-old quantum computing problem that clarifies the limits of quantum speed advantages over classical computers. Jensen Huang resigned as Nvidia CEO after 33 years following reports of affairs with an alleged Chinese spy and a company employee, with CFO Colette Kress stepping in as interim CEO at the world’s most valuable company ahead of its November 17th earnings announcement.
|
| |
|
| | | #1264
Sunday, Someday. | | | | |
|
