A teacher wanted three bullet points on what made a ten-year-old special. Marissa Mayer asked Dazzle, her personal AI assistant. It came back with “empathetic, creative and an animal lover.” She told the New York Times the answer was perfect.
Alexandra Alter’s profile, published October 25, 2026, frames that scene as a founder getting unstuck. The Times described Mayer as a former Yahoo chief executive and Google’s first female engineer, with a startup team of 10 engineers. She was overwhelmed. The software helped. She liked the result.
On TechCrunch’s Equity podcast, the same scene read as a status story. A person with engineers on staff, and an assistant made of software, had handed off the one description a teacher had asked a parent to write.
On July 31, 2026, Sam Altman offered a cousin of the idea. Connect the family calendars, describe what the kids care about, and wake up to a generated podcast: one child’s soccer game, another’s upcoming birthday, a little news. A Business Insider column said the backlash was swift, and that it boiled down to a simpler proposal. What if you just talked to your children?
The wrong kind of friction
These stories travel because they are intimate, not because anyone has published a study of parents who want a model to narrate the school run. Kirsten Korosec asked on Equity whether the disconnect was a marketing problem or a product problem. Amanda Silberling’s answer was short: “A little bit of both.”
The pitch sounds far from ordinary life, and the products often chase the wrong inconvenience. Silberling’s example was her own week. She is not booking flights all the time, and a dinner reservation is not the hardship that organizes her week. Anthony Ha argued that the executives making these pitches are paid to claim the technology belongs in every corner of a life. Push that claim hard enough, he said, and the picture stops looking appealing. Sometimes it looks dystopian.
Silberling was willing to sound unfashionable about it.
Call me old school, but I think that humans being able to talk to each other is probably a net positive for society.
Easy to applaud, and too broad if you stop there. A Business Insider column drew a cleaner line than “keep the machines out of the family.” The writer would happily hand off diapers, laundry, camp signup, and the tangle of school messaging apps. Not the conversation in the car, and not a bedtime story made up on the spot. One is drudgery. The other is a reason people have children. That is an opinion, not a survey, and it is a more useful product test than a blanket no.
The useful version already exists, and it is boring on purpose. Business Insider, via Ben Shimkus, described a Ford executive who built a family “chief of staff” that emails both parents a daily rundown of drop-offs and activities. Nobody in that setup pretends the software knows the child. It sends both adults the plan. That is coordination. A generated podcast about the soccer game is a substitute for asking, from the driver’s seat, how the game went.
Ha does not buy the industry’s usual sequel. In that story, early adopters are just early versions of everyone else, the way Uber escaped the young, well-off San Francisco users he once thought would be its whole market. He does not see that sequel yet for AI that wants to parent or narrate a life. The pushback he described was plainer. People want to live their own lives, and they want to parent their own children.
Treat that as an argument, not a census. The podcast brought no adoption figures. Silberling did not pretend to know what a typical person asks ChatGPT or Claude on a Tuesday. Her claim was narrower: the value people are finding is not “book my flight.” Elite inconvenience is a bad compass. The alternative still has no measured map.
What a buyer should actually automate
Using AI is not the executive mistake. Mayer was overwhelmed, the assistant helped, and she liked the output. All of that can be true while the example still fails as a picture of everyone else’s problem. The miss is generalizing from a life in which the scarce resource is attention, then selling total life integration as if the rest of the country is stuck on the same pain.
The practical test is simple. AI automation earns its keep when it takes a recurring, disliked, checkable task and returns something a person can accept or reject.
On a sales team, that is an AI agent that pulls the last three calls, flags the open objection, and drafts the follow-up a human sends. That is the sensible shape of AI for sales: the agent prepares, the person still shows up. In support, it is triage and a drafted reply someone approves before it goes out. In finance, it is an agent that flags invoice exceptions rather than paying the wrong vendor. It is not a system that takes the customer meeting so the team never has to learn the account.
Silberling’s line on Equity belongs next to that test: “we haven’t found the balance of what optimization actually helps us and what optimization just makes us feel like we aren’t as human as we want to be.”
If you want to see where money is already following a version of that test, look at contracts, not carpools.
Where the contracts already are
Rebecca Bellan reported in TechCrunch on October 26, 2026, that Anthropic had filed for an IPO and planned to list as ANT. The amount to be raised and the proposed share price were still undisclosed. The operating picture, as TechCrunch drew it from the filing, was more specific.
In the first half of 2026, revenue was $16.3 billion, up from $1.2 billion in the first half of 2025, more than thirteen times the year-earlier figure. The operating loss narrowed to $693 million, from $2.1 billion in the first half of 2025. The net loss was $80.7 billion. TechCrunch reported that figure as largely driven by noncash accounting adjustments tied to convertible debt, not cash out the door. Those are different facts. A headline that treats the net loss like a cash bonfire is how a filing starts to sound like collapse.
The company also said it may not achieve or sustain profitability. As of the end of September 2026, according to that same report of the filing, more than 11, 900 customers were spending $100, 000 or more a year, 1, 900 were spending $1 million or more, and about 250 were spending $10 million or more.
That is AI for business, sold as infrastructure. This is Anthropic’s book. It is not a census of parents paying for a better description of a child, and it is not proof that nobody is. One company’s annual contracts show where that company’s serious money sits. They do not settle whether people will pay for intimate tools. This debate still has not measured that.
The industry will keep offering a smoother life. Some of that smoothness is a morning email both parents can trust. Some of it is a podcast standing in for a question from the driver’s seat. The companies that last will be able to tell those apart before their customers have to.
Questions worth asking
- Did the Times scold Mayer for asking an AI to describe her daughter?
No. Alter’s profile treated Dazzle’s three traits as a win for a busy founder. The harsher reading comes from later commentary, including Equity, not from the profile itself.
- Is the backlash a rejection of any AI help with family life?
Not in the sharper version. A Business Insider column argued for handing off logistics and keeping the car conversation and the bedtime story. That is opinion, not a survey, and it is a better line than a blanket no.
- Where is the documented demand?
In the enterprise book TechCrunch described from Anthropic’s filing, not in a usage study of parenting podcasts. As of the end of September 2026, those reported figures included thousands of customers above $100, 000 a year and about 250 above $10 million. The absence of comparable consumer numbers is not proof that intimate demand is zero.
- Does the huge net loss mean the AI business is unraveling?
Not by itself. TechCrunch’s account of the filing showed first-half 2026 revenue rising from $1.2 billion to $16.3 billion and a narrower operating loss, while the $80.7 billion net loss was reported as largely noncash and tied to convertible-debt accounting. The company still warned it may never be sustainably profitable.
- Should you buy the pitch that AI belongs in every part of the work, and of life?
As a buying rule, not a survey result: pay for automation that removes work people resent and that a person can check. Leave the conversation, the judgment, and the relationship with a human. A product that has to occupy every hour of a customer’s life is easy to pitch and easy to resent.