Dylan Keilreads the ocean's mind.

An infomercial oven, and the strangest career in shipping

Dylan Keil was a kid mesmerized by the “set it and forget it” ovens spinning on late-night infomercials. Not the chicken. The promise. A machine you could trust to do the right thing while you walked away. Three decades later he runs a company that does exactly that for 50,000-ton ships crossing oceans you will never see.

That company is Bearing AI, and Keil is its co-founder and chief executive. The pitch is deceptively simple: feed an artificial-intelligence model enough data about the world's cargo fleet and it learns to predict how fast a given vessel will go, how much fuel it will swallow, and how much carbon it will leave hanging in the air - all without a single sensor bolted to the hull. No onboard hardware. No IT department wrestling with integrations. The model already knows your ship, because it has studied thousands of ships just like it.

Shipping is the invisible circulatory system of the global economy. Roughly 90% of everything you own arrived by sea, and the boxes that carried it ran on heavy fuel oil thick enough to chew. The industry produces close to 3% of the planet's carbon emissions - put it on the map and it would rank among the top ten countries. For most of history, the people steering those ships flew half-blind on instinct, weather faxes, and the captain's gut. Keil's bet is that a well-trained model can see what the gut cannot.

Growing up in the dining hall

He did not arrive here by way of the docks. Keil spent the first 22 years of his life on college campuses - both his parents were academics - which meant childhood dinners were a rotating seminar. Physicists, historians, economists, all arguing across the same table. He absorbed a habit there that would define him later: the generalist's refusal to stay in one lane. “Some of the biggest problems we face in the world are physical,” he has said. “Climate change, energy, water access.” The instinct to chase the physical, not the purely digital, runs straight back to those campus dinners.

He went to Stanford for mechanical engineering, certain he would become an engineer who built things you could hold. Somewhere along the way the definition stretched. “Making something,” he realized, “also includes building a business.” He stayed at Stanford for an MBA, then went looking for the seam between the digital and the physical worlds - the place where software touches steel.

Sensors, phones, and a detour through family safety

His first landing spot was an early-stage startup riding the launch of the iPhone, working with sensors strapped to mobile devices. The intersection of bits and atoms again. In 2012 he co-founded Chronos Mobile Technologies, a contextual-awareness engine that read a phone's location data to help people set and hit time-based goals. Chronos worked well enough that Life360 - the family-tracking app - bought it in 2015.

Keil stayed on as Director of Product and helped scale Life360's flagship app past 30 million users. It was a masterclass in turning raw sensor data into something a normal human would actually use. But family safety, however large, was not the physical, planet-scale problem he kept circling back to. In 2019 he left to find one.

The problem found him in an unglamorous corner of the world economy: a fleet of enormous metal boxes burning the dirtiest fuel on Earth, run by an industry that had barely been touched by modern machine learning. He co-founded Bearing AI with the backing of the AI Fund, the venture studio built by Andrew Ng - the man behind Google Brain and Coursera, and one of the most quoted names in the field. Ng took a board seat. Japanese trading giant Mitsui & Co. came in alongside him. In 2022, Bearing raised $7 million in a post-seed round, bringing its total to about $10 million.

The report card for the ocean

Bearing's timing turned out to be uncanny. In 2023, new rules forced every shipowner to report a Carbon Intensity Indicator - a CII score - for each vessel, graded on a scale from A to E. Think of it as a fuel-economy report card for ships, with real money and reputation riding on the letter. The catch: owners had no reliable way to know what grade they would get, or how to nudge an E toward a B. The industry's blunt instinct was to simply slow every ship down.

Keil offered a sharper tool. Bearing launched an AI simulator that predicts CII scores for every vessel afloat, then models the levers that move them - speed, fuel type, sailing schedule, even when to scrape the barnacles off the hull. “Companies know the new CII regulation will affect their business,” he said at the launch, “but they have no way to accurately assess its impact on a ship-by-ship basis. Our CII Optimization product permits shippers to see what we see and take the steps they need to reduce their environmental impact while improving their bottom line.”

That last clause is the whole philosophy compressed into a sentence. Keil does not ask shipowners to choose between the planet and the profit-and-loss statement. He insists the two point the same direction. Burn less fuel and you spend less money and you emit less carbon - the rare problem where the green answer and the greedy answer are the same answer. “Success and happiness are not mutually exclusive,” he likes to say. In maritime form: doing good and doing well need not be either.

Why it matters, in numbers

By 2023, ships sailing on Bearing's guidance had logged more than 75 million miles - roughly 3,000 trips around the planet - and saved over a million tons of carbon along the way. Andrew Ng put the company in front of his enormous following to make the point: “While AI can't solve climate change, it will be a key part of the solution.” Keil's own math is bolder and simpler. Shave global shipping emissions by just 10% - which he calls genuinely achievable - and you have erased the equivalent of millions of cars from the world's roads. Not by inventing a new fuel. Just by sailing smarter.

Bearing has since pushed past the single ship. Its Fleet Deployment Optimizer plays chess with an entire fleet at once, deciding which vessel should carry which cargo where, and the platform now reaches into Europe's tightening web of rules - the EU Emissions Trading System, FuelEU Maritime - where a wrong move costs real euros. Today the company describes itself as “maritime intelligence built for the era of LLMs,” a phrase that would have meant nothing to the shipping world a few years ago and means everything now.

Keil's style cuts against the Silicon Valley grain. He is wary of flash, partial to plumbing. Effective AI, he argues, should hand a captain an actionable decision, not a beautiful dashboard - insight that hits the bottom line in real time, not a chart to admire. It is the “set it and forget it” instinct, grown up and gone to sea. Build the thing that quietly does the right thing while everyone else is busy. Then walk away and let it work.

For someone chasing the most physical problems he could find, there is a pleasing symmetry to where he landed. The ocean is about as physical as it gets - 360 million square miles of it, indifferent and enormous. Dylan Keil decided the way to move it was to teach a machine to read it first.