At some point in crafting a complex system, says Smartest Systems founder and CEO Julian Loren, most people throw up their hands. “They say ‘it’d be nice if we could see all this in one picture but we can’t.” Actually, he says, they can — and that he thinks he’s figured out how.
I ran into him at a monthly gathering in San Francisco for sustainability-oriented startup people. Loren’s on the prowl, he says, for opportunities for what sounds to me — admittedly an AI and supply chain novice — like interesting technology.
He calls his innovation “AI ensemble management” — a fancy name that seems to live up to his tech. He and others have been working on it since before Loren left GE four years ago, since the mid ’90s has worked on artificial intelligence.
He’s proud to tell about a tryout of his technology at the Port of Los Angeles. For one week, after two weeks’ lead time, he says, “They broke every single record.” The port moved more containers per longshoreman, and it moved more containers per terminal truck, among other outstanding records it set that week. “By any metric, they broke all records for unloading the ship because they had that kind of foresight.”
Typically, he explains, a ship that takes weeks to cross the Pacific transmits its cargo manifest only 12 hours ahead of docking. In that final 12 hours, the port scrambles to line up labor and equipment for unloading.
His systems descend from early, handcrafted cause-and-effect models built by “a bunch of very smart guys doing a lot of detective work.” Later came dynamic, just in time models. The current generation, embodied in Smartest Systems,
Eventually, Smartest Systems developed its “inference engines” produced by “AI ensembles.” Instead of relying on just one AI technology, they team up multiple AI technologies and learned how to weight each one’s results. These let them predict a context, then use that not-yet-existing context to create simulation models and run what-ifs.
“If I have a ship sailing from China,” he says, “I want to know the most likely context when the ship arrives.” Which straddle cranes will be in for repair, how many chassis will be there to put containers on, how many semi tractors to pull containers? “If I can do that, I can run what ifs. See where the gaps will be to act beforehand.” He has time to prepare.
“Suddenly, they’re not reacting in chaos and touching containers a whole lot more than you have to,” he says of the Port of Los Angeles trial. “It was completely choreographed.”
“Basically,” he says, “you can turn the industrial and commercial world into a super-choreographed ballet instead of the run-around-with-your-head-cut-off fire drill that it is now.”
But wait, there’s more to it! This month he’s convening a group of clients near Los Angeles to orchestrate the human ecosystems that aid Smartest Systems. I hope to have some account of that here eventually.