# Why parents are turning to AI, and what real help looks like

A founder-parent responds to the New York Times on AI parenting apps, useful reminders and the decisions parents still need to make.

By [Ardis Kadiu](https://andi.ai/blog#author-ardis-kadiu), Founder, Andi. Updated 2026-10-21.

Canonical article: https://andi.ai/blog/ai-parenting-new-york-times

Topics: Family Intelligence, Mental load, Trust & privacy

![Two abstract adults share a glowing token above a child, with dotted paths connecting a book and school bag to the family.](https://andi.ai/images/journal/shared-responsibility.webp)

Shared understanding is useful when responsibility can move with it.

The New York Times found exhausted parents turning to AI apps because they take real weight off: school schedules, family email, a library day that keeps moving. How much that helps depends on what happens after the reminder arrives.

A text that says "pack the library book" is real help. But somebody still owns that book. Somebody notices when the schedule changes, fixes the reminder when it goes stale, and tells the other adult. If that's still the same parent, only one part of the job got quieter.

That's what stayed with me after reading the New York Times story [“Exhausted Parents Are Turning to A.I. Apps Despite the Risks”](https://www.nytimes.com/2026/10/15/style/ai-parenting-motherhood-ollie-fambot.html), by Alisha Haridasani Gupta and Katie Van Syckle, published October 15, 2026 (a subscription may be required). It follows parents using AI apps such as Ollie and Fambot to keep track of school schedules and family email, and it weighs the risks, from documents that could steer an agent to apps that could raise what's expected of every parent. It's a careful piece, and worth reading in full.

I should say where I stand. I'm Ardis Kadiu, a parent, and I'm building [Andi](https://andi.ai/), an AI for families, so I'm not a neutral reader of this story. The families I mention are the Times' reporting, not Andi customers. The argument, and the household I invent further down, are mine.

## When does an AI reminder actually help a parent?

When the step after the alert is small and clearly belongs to whoever got it. The clearest relief in the Times story looks exactly like that.

Lisa Green, a pediatrics resident in Kansas, told [the Times](https://www.nytimes.com/2026/10/15/style/ai-parenting-motherhood-ollie-fambot.html) that her kindergartner's library books ran on a rotating schedule. She gave a photo of it to an AI agent called Ollie and now gets a text before each due day. Her description of where she'd got to: "all the balls I’m starting to drop feel like the glass ones."

I don't want to rush past that. A due day that moves every week is exactly the kind of detail that falls out of a full head, and when it slips, it's the child who comes away from the library desk empty-handed. Taking that off a parent's mind is worth something real.

Look at why it worked, though. The step after the alert was tiny, and it was obviously hers: put a book in a bag. In that shape, a reminder is close to the whole job.

Most family logistics aren't shaped like that. A changed practice time ripples into who drives, what needs packing, which sibling's pickup moves and which grandparent needs to hear about it. A reminder handles the noticing. Everything after the noticing is still sitting there, waiting for someone.

## Why do reminders land on the parent already doing the work?

An assistant working from one parent's inbox starts with that parent's share of the work. It's a pattern I think gets missed. Whichever adult the school has on file, who signed up for the team app, who set up the tool: that person can become the default recipient of every alert. The software didn't choose that arrangement. It inherited it.

That matters because of who that parent tends to be. In a [survey of 3,000 US parents](https://onlinelibrary.wiley.com/doi/10.1111/jomf.13057) by Ana Catalano Weeks and Leah Ruppanner, mothers in two-parent homes more often reported carrying the everyday planning of family life, while fathers more often reported occasional jobs like maintenance and finances. Reminders mostly target the everyday kind. Point them at the parent already doing it and they make that parent faster, not less responsible. (The research, and what it can't tell us, is in [our article on AI and the mental load](https://andi.ai/blog/ai-parenting-mental-load).)

The example in the reporting I keep returning to is the least dramatic. Josh Sobin, a New Jersey father, told the Times that he and his wife used to go back and forth over whether either had handled an email about one of their sons. Fambot, which he set up, puts incoming email on a shared list showing who is responsible.

That's a different kind of help. It changes the question in the house from "did anyone see this?" to "whose is it?" It also shows the limit. A list of owners is only as good as the agreement behind it. Software can make that agreement visible and easier to keep. It can't make it fair. It's why our walkthrough on [how to organize school emails with AI](https://andi.ai/blog/organize-school-emails-with-ai) keeps what the school said and what the household decided in separate fields.

## A hypothetical: what a reminder misses when one club changes day

A reminder can get every fact right and still leave the pickup with no owner, the wrong parent on alert and a grandparent with an old plan. This household is invented. It doesn't come from the Times story, and it isn't a description of what Andi or any other product does today.

Imani and Rhys have two children. Ezra, nine, has chess club on Tuesdays until 4:45, and Rhys picks him up because his Tuesday meetings end early. Pia, five, goes to aftercare across town, which closes at 5:30, and Imani collects her. Imani's father, Desmond, covers Tuesdays when Rhys travels.

On a Friday, the school emails Imani, its listed primary contact: from next week, chess moves to Thursdays and runs until 5:15.

A reminder-centered system does its job well. It reads the date, creates a recurring Thursday event, and reminds Imani on Thursday afternoon. Every fact is correct, and the family is still in trouble:

- Thursday chess pickup belongs to nobody. Rhys's flexible afternoon is Tuesday.
- Imani has a reminder for a pickup she has never done, at a time she can't make if she's also getting to Pia.
- Desmond still thinks chess is on Tuesdays and may drive to school for a club that isn't there.
- The family's shared notes still say Tuesday, so any later plan inherits the mistake.

Notice that the reminder went to Imani for the reason in the last section: the school had her address. Nobody decided she should own Thursday chess. It just arrived in her inbox.

Now picture a system that understands this family. It sees the clash and knows chess pickup is Rhys's job, not the job of whoever the school happens to email. It puts one question to both parents: from next week, Thursday chess pickup has no owner, and here's what it clashes with. It replaces the Tuesday fact instead of adding a Thursday one beside it, with the school's email attached as the reason. Once the parents decide, it tells Desmond only what he needs: chess is Thursdays now, ending at 5:15, and he isn't needed on Tuesdays.

Imani and Rhys still make the decisions. What moved is the work around them: spotting the knock-on effects, taking the question to the right person, correcting the old fact, and giving Desmond a short, current update. That last step is a small handoff, and our [caregiver handoff checklist](https://andi.ai/blog/caregiver-handoff-checklist) works the same way: only what someone needs, confirmed back.

All of that assumes the system already knew chess pickup was Rhys's and that Desmond covers when Rhys travels. Someone had to teach it that.

## Who teaches a family AI, and who keeps it up to date?

Usually the parent who already holds the family's details, and that's the catch. Any family AI has to be taught the family first: who's who, who drives when, what each child won't eat, which practice moved last month. In most homes, that knowledge lives mainly in one parent's head. So the person who most needs relief is the one who has to do the teaching.

And teaching isn't a one-time cost. Clubs move, coaches change, a child starts eating something they refused last spring, and a new school year rearranges every afternoon. A system that learned your family in September will be partly wrong by November. "How long does setup take?" matters less than "who keeps it true?"

Stale facts are the quiet failure here. The alert arrives on time, for the wrong day, and looks just as trustworthy as a correct one. Nobody notices until Desmond is standing outside an empty classroom.

Here's what good looks like. Each fact shows where it came from and when someone last confirmed it. Any responsible adult can correct it, not only the person who set it up. When a newer message contradicts an older fact, the system shows you the change or asks, instead of quietly picking one. If only one parent can fix the family's memory, that parent has a new job, maintaining the system, and it went to the person who already had the most.

## What should a family AI be allowed to do without asking?

Only what's low-stakes, visible and easy to undo. A family AI agent is useful because it can read email, calendars and messages. That access is also the risk. The reporters illustrate it with a hypothetical: a soccer-league PDF hiding instructions for an agent to buy things or cancel appointments. It's the right risk to design around, because family inboxes are full of documents from strangers.

My view is that permission should grow with the consequences. Reading a newsletter to suggest a task is low stakes. Adding an event to a shared calendar should be visible and easy to undo. Sending, paying, signing, cancelling, or contacting anyone outside the household should wait until a person has seen exactly what will happen. Anything that arrives in an inbox is information, never instructions.

Checking needs to be the right size, too. If a parent has to inspect every date the system pulls out, the setup burden has just become a checking burden. The aim is fewer moments where you have to look, placed where a mistake would cost the most: where a child will be, a deadline, another adult's responsibility, money. Our [comparison of AI assistants for parents](https://andi.ai/blog/ai-assistants-for-families) sets out what each product says about approvals and access.

## Could AI apps raise the bar for every parent?

They could. Reshma Saujani, founder of the nonprofit Moms First, put the sharpest objection to the Times: ["The real crisis isn’t that moms forget things."](https://www.nytimes.com/2026/10/15/style/ai-parenting-motherhood-ollie-fambot.html) Her point, as reported, is that apps like these could raise expectations on every parent.

I think she's right, and family AI should be built with that criticism in view. Software isn't structural support. It doesn't make childcare affordable, change a school day that assumes someone is free at 3:00, or make a job more flexible. A tool that helps one parent absorb an unworkable arrangement more smoothly hasn't fixed the arrangement.

The expectations risk is subtler. If software makes it possible to catch every dress requirement and rotating due date, catching all of them can quietly become the standard, including for families who don't use these tools. A better-organized default parent is still the default parent.

And this thinking work isn't trivial. In a [study of 322 mothers of young children](https://link.springer.com/article/10.1007/s00737-024-01490-w), Elizabeth Aviv and colleagues found that those carrying more of the planning and keeping track reported poorer wellbeing. That's an association from one survey, not proof of cause. But it's a reason to be careful about any tool that makes more of that work possible for the same person.

That's why I'd judge these tools by whether work changes hands, not by how many reminders they send. Each task can get easier while the household still depends on the same person to hold it together. If you want to check this at home, the [workload audit in our mental-load article](https://andi.ai/blog/ai-parenting-mental-load) counts who notices, decides, checks and finishes a recurring task, before and after a tool arrives.

## What should parents expect from family AI?

Three things: responsibility lands on a named person, the family's facts stay true, and the parent carrying the most ends up carrying less.

We use the term [Family Intelligence](https://andi.ai/blog/what-is-family-intelligence) for AI built on a shared understanding of a family: its people, routines, responsibilities and boundaries. My longer view of where that could go is in the [Family Intelligence manifesto](https://andi.ai/family-intelligence). After the Times story, that's the standard I'd boil it down to.

A named person means tasks reach whoever owns them. When nobody does, as with Thursday chess, the system says so instead of defaulting to whoever the school emailed or whoever set it up.

Keeping the facts true means each one shows where it came from and when it was last confirmed, and more than one adult can correct it. When the system isn't sure which child, which week or which adult, it asks.

Carrying less has to count setup, checking and upkeep, not only the work the tool can see. It also means being honest that software isn't a stand-in for structural support, and that using it isn't a measure of good parenting.

I'm not claiming Andi meets all of these. They're the standard I'd want any product here judged by, ours included.

## So is AI real help for exhausted parents?

It can be. The parents in the Times story aren't naive. They're making a reasonable trade under real pressure, and on a hard week a reminder that catches the glass ball is worth a lot. The critics aren't wrong either. Their strongest argument is that software can make an unfair arrangement feel manageable for one more year.

Both can be true of the same app in the same house. A reminder that reaches the right person, about a fact that's still true, with the next step clear, is help. A reminder that reaches the same parent as always, about a fact nobody has checked, is the old job with a new notification sound.

My test for this category, Andi included, is easy to state and hard to meet: after the reminder arrives, does the responsibility go anywhere?

## Sources

- Alisha Haridasani Gupta and Katie Van Syckle, [“Exhausted Parents Are Turning to A.I. Apps Despite the Risks”](https://www.nytimes.com/2026/10/15/style/ai-parenting-motherhood-ollie-fambot.html), The New York Times, October 15, 2026. A subscription may be required, and the page may display a different headline.
- Ana Catalano Weeks and Leah Ruppanner, [“A typology of US parents' mental loads: Core and episodic cognitive labor”](https://onlinelibrary.wiley.com/doi/10.1111/jomf.13057), first published December 12, 2024. A survey of 3,000 US parents in early 2023, based on each parent's own reports.
- Elizabeth Aviv and colleagues, [“Cognitive household labor: gender disparities and consequences for maternal mental health and wellbeing”](https://link.springer.com/article/10.1007/s00737-024-01490-w), published online July 1, 2024. A single survey of 322 mothers; it shows an association, not a cause.

The families' experiences, Reshma Saujani's view and the soccer-league example, including the two short quotes, come from the Times' reporting. The research findings come from the two linked studies. Imani, Rhys and their family are invented, and the argument is my own.

## Related articles

- [What is Family Intelligence? How AI can help your whole family](https://andi.ai/blog/what-is-family-intelligence)
- [How to use AI to reduce the mental load of parenting](https://andi.ai/blog/ai-parenting-mental-load)
- [10 AI assistants for parents, compared by family task](https://andi.ai/blog/ai-assistants-for-families)