The product Peter Diamandis described on the Moonshots panel fits in one sentence. You fill out a values questionnaire. Every Wednesday at 7 p.m., a text arrives with a match, plus a place and a time to meet. "That's the entire product." No feed, no swiping, no infinite scroll — you show up and see whether anything happens.
Ditto, a Berkeley startup, describes the same workflow on its product page: an introduction delivered by iMessage on Wednesday evening, and, once both people reply, a date with the scheduling handled for them. The company says the service learns users' preferences and that more than 80,000 dates have been arranged. On the episode, Peter put the figures at 160,000 college students signed up and 80,000 dates produced. That number counts meetings the service coordinated, not relationships that lasted.
Peter's framing was that this is the old matchmaker's job, "a yenta, if you would," rebuilt as software. The dating industry assumes more options are better; Ditto's bet is that too many options produce decision fatigue and analysis paralysis, and that the fix is to hand the choosing to the machine.
Deleting the interface
Salim's first reaction was about the shape of the product rather than romance. Tinder, he said, optimized searching — better filters, better corrections, more sorting. Ditto makes the searching unnecessary. What interested him is that the AI here is not adding an interface but deleting one: the user's job shrinks to showing up and finding out whether there is chemistry, which he noted you have to do in person anyway.
He expects the same pattern to move well beyond dating, and said he is looking at building something like it for business connections. Later in the conversation he returned to Peter's yenta line and predicted the Indian matchmaking industry would be disrupted by the same mechanism, since the requirements families specify — caste, clothing, values — are exactly the kind of thing you can write down and hand to a matching system. The broader pattern, as he put it, is AI becoming "the trusted intermediary between individuals interfacing with overwhelming abundance." The question that leaves open, in his words: do you want to outsource that trust?
Peter gave a concrete version of the same problem from his own events. At the Abundance Summit, matching entrepreneurs among 600 CEOs is one of the most important things the organizers do, and, he said, randomly bumping into the right person among 600 people over five days is tough. That, to him, is where a matching system earns its keep.
The salesperson problem
Dave said he would have backed the company without hesitation, and he sees it as a stepping stone to AI managing your choices in general. Done right, he said, that would be "one of the greatest boons to mental health in world history." Done badly, it would be horrible — "if it's left to manipulate you" — because such a system is such a great salesperson. Dating is a clean test case for him: is the thing trying to lead you to the right person, or selling you something you don't want? He has long argued that the advertising model disappears once your AI knows you well enough to simply buy what you need; the same intimacy is what makes the failure mode ugly.
Emad's worry ran alongside that one. He recalled a Black Mirror episode in which people send digital twins on dates to test compatibility in milliseconds, and said the direction feels similar. His concern is the point where "you can't argue with your AI. It knows best." He invoked the phrase "all watched over by machines of loving grace," and said that outsourcing your cognition and connection "does take away a little bit from your intrinsic humanity." He said he already sees a version of it at work: colleagues doing good work who slip into trusting the AI too much, to the point of sending something machine-written as though it were their own.
The one panelist who said no
Asked whether he would try it as the only unmarried person on the panel, Alex said no: "this is why we can't have nice things." His objection was not to the Wednesday introductions but to another tool he said Ditto released — what he called the Ditto AI body count detector, which he described, quoting what he said was the company's own website, as using 478 facial points and 52 micro-expressions over five seconds to estimate how many sexual partners a person has had. No one else on the panel had heard of it; Peter asked him to explain. It is Alex's account of a separate feature, and it does not appear in the matchmaking workflow Ditto documents.
His complaint was about allocation: "a politely suboptimal use of scarce reasoning tokens," when, as he and Peter had written elsewhere, the same compute could go toward solving everything. Social discovery is valuable to him when it serves socially or economically productive purposes; body count detection reminded him of Hot or Not in the early Facebook days. "We could be aiming so much higher as a civilization."
Peter defended the underlying idea. Most people on dating apps, he said, judge on external parameters — handsome, beautiful — while matchmaking works past the initial hormonal response, and some of the longest-lasting marriages across cultures have come from being matched. He added two conditions of his own: how honest you are on the questionnaire matters, and whether the service has enough data to match people accurately "is a different thing." With the US divorce rate at 50%, he argued, a system that genuinely found the best match would carry massive societal value, even if the parameters in use today are the wrong ones.
Alex held his line while correcting the wording: "Not a waste, a suboptimal use." His reconciliation was a scheduling one. Once there is a leisure class and tokens too cheap to meter — which, he said, does not exist yet — do as much body count detection as you like. Dave agreed on the ordering: after the major diseases are solved is probably a good time to start. Peter moved the show on, saying he hopes Ditto works and that many happy relationships come out of it.