Keith Rabois, Travis Kalanick: Grok 4 Wows, The Bitter Lesson, Elon's Third Party, AI Browsers, SCOTUS backs POTUS on RIFs

The episode opens with joking about a stay at a hotel on Lake Como. One host says they left a message in the hotel's registry book and took home the hangers, laundry bags, robes and slippers. The host then introduces the week's panel. It is summer, some regulars are busy, and two guests fill in. One is introduced as "another brilliant PayPal alumni and, dare I say, GOP supporter," joining from New York. The other is introduced with the line "Travis Kalanick is back with us," on his second appearance at the roundtable and his third on the show.

While the New York guest is introduced, the panel briefly discusses the city. The guest says the city is safe for now and says July 4 was "the first time in recorded history that there were no shootings or no murders in New York on that day." The guest adds, "We'll see on November 4," and says "we may be leaving New York quickly." The host then jokes that a mayoral candidate, transcribed as "Mondami," would seize property.

A panelist mentions Cloud Kitchens and says "we're rolling out robots." The panel then asks about reported news involving Pony AI and calls it "speculation." A host gives the background. Pony AI is a self-driving company based in China, one of the few with cars on the road, with large operations in the Middle East and a deal with Uber. The reported idea was to partner with Uber, license Pony's technology, and start a competitor to Waymo and Tesla.

The answer begins with the US market. Waymo runs in San Francisco, Los Angeles and Austin, with Miami, Atlanta and DC described as coming soon and New York under discussion. Tesla is described as "doing it the hard way, you know, classic Elon style," and it is "unclear when it gets over the line," despite a recent semi-pilot in Austin. The speaker's point is that "there's no other alternatives." People who want alternatives have reached out, and the speaker reports being told, "Travis, you did autonomy way back in the day."

The speaker links this to food. Automating both the kitchen and food logistics would "take huge amount of costs out" of food. The speaker argues that solving autonomy once could cover moving things, food and people. The speaker is careful about the status: "There's no there's no real deal right now, but there is definitely some inbound." The speaker also says there is news about "some of that inbound that may or may not be occurring."

The panel turns to a bowl-building machine, which a host thinks is called Lab 37. The panelist describing it says it is 60 square feet, although the picture "makes it look monstrous." The machine can run several brands at once, for example a Sweetgreen-like or Chipotle-like brand. A customer orders a custom bowl online. Small carriers move the bowl under dispensers. The machine sauces the bowl, puts a lid on it, bags it with utensils and seals the bag. A conveyor and an AGV then take it to a locker at the front of the house. A courier opens the right locker by waving an app in front of a camera.

The claimed benefits are accuracy and cost. "We know exactly how many grams of every ingredient are put in," the speaker says, which makes mistakes hard and gives a higher-quality product. Customers are starting to deploy the machine this quarter. In the company's delivery kitchens, labor costs about 30% of revenue, or 30% to 35% for successful operators, and more in brick-and-mortar restaurants. With the machine, the transcript gives the figure as "between 710% of revenue," which is most likely a garbled "7–10%."

The speaker compares this to Uber's original tagline, "Everyone's private driver." For ride-hailing, the roads and cars already existed, and people "weren't using their cars 98% of the day." Restaurants do have spare capacity, which Uber Eats uses. But preparing and delivering 30% of a city's meals would need new infrastructure: "the infrastructure is not there, so you have to build it." The company's stated mission is "infrastructure for better food," meaning real estate, software and robotics for producing and delivering food.

A panelist asks whether this is what David Friedberg tried a few years ago, a reference to the company Itza. The panel jokes that the business "died as a vegan martyr." The panelist claims people did not want to eat quinoa and that Friedberg eventually added protein. The joke is aimed at the menu. The more serious explanation of why such efforts fail comes next.

The argument is that food automation has to be end to end. The speaker cites automated pizza companies that "have come and gone." In those setups, one worker feeds ingredients into the machine and another boxes the pizza: "So instead of one guy making pizzas, I have a million dollar machine and two guys making pizza." In the speaker's model, staff prep the food, load the machine and leave, and the restaurant "runs itself for many hours without anybody there."

The speaker separates two stages. Assembly, also called the make line, is putting a meal together and plating it. "State change" is the cooking. For a Chipotle or Sweetgreen, assembly is described as the majority of the labor. The speaker says a lunchtime Chipotle may have "10 guys" working while customers still wait in line, and that the machine does 300 bowls an hour. Over time, production lines for things like cooked burgers would feed into the assembly line from the side. One participant compares the result to "factorio on steroids."

A panelist asks what this means for real estate. If one machine makes 300 bowls an hour, a small faceless warehouse with five machines and drone or car delivery could replace storefronts. Would that leave "a wasteland of real estate"? The reply is that about 85% of US meals are eaten at home, most of them home-cooked. Uber Eats and DoorDash account for "1.8% or 2% of all meals right now." On this view, the service replaces home cooking more than it replaces restaurants. People will still go out, which "we learned that during COVID." The speaker summarizes the aim as "you don't have to be wealthy to be healthy."

The speaker looks further ahead: "in twenty years, I think everybody agrees you will have machines making very high quality, very personalized meals for everybody." Another participant suggests the strongest version may be a robot private chef in every home, because people want to cook at home but lack the time. A host notes that wealthy people already pay for meal services costing $200 a day. The host argues empty commercial space gets reused, citing storefront live-work spaces in 1980s and 1990s New York and malls turned into colleges and artists' studio apartments. The host also says the country still needs "five, ten million homes."