this robot can fold your laundry

here's your tl;dr:

you need this robot as much as you need a roomba

so, you might need it 🤷🏽‍♀️

I hate folding my laundry.

Honestly, I hate everything about doing laundry—from remembering to put it in the washer, to moving it to the dryer, to inevitably leaving it on my bed where I either forget to fold it and put it away, only to have to dig through the pile the next day for leggings to work out in.

I don’t know how you feel about doing laundry, but I’d imagine I’m not alone in this complaint — particularly if you have children, or otherwise need to do a lot of laundry each week.

In fact, I know I’m not alone, because this complaint goes viral every few months — see: The Folding of the Laundry trend on last year, any online conversation about fitted sheets, etc. And with that trend often comes the mention that it would be great if AI could fold laundry so that we don’t have to.

Well, it turns out we might be closer to an AI laundry-folding reality than you may have thought.

A Robot That Folds Laundry (among other things)

As I was scrolling Twitter yesterday, I came across a video of a robot folding laundry.

This video was part of a larger announcement yesterday from Physical Intelligence, a San-Francisco-based startup that focuses on “developing foundation models and learning algorithms to power the robots of today and the physically-actuated devices of the future.” Their recent blog post details the development of π₀, their first prototype general-purpose foundation model designed to make progress towards this goal.

More importantly, they released a bunch of fun videos demonstrating the model’s use on tasks including: folding laundry, opening popcorn, bagging groceries, bussing tables, and more.

I think this is pretty cool. It's a step toward what many of us have quietly wished for: tools that handle the mundane tasks and give us time to focus on the pursuits we care about.

And I think that’s why I was a little bit surprised at how little attention this got.

It’s not actually that surprising — there were a ton of other announcements almost every day this week (see also: every other week) from much larger companies, so cutting through that noise would be a challenge for pretty much anyone. Yet without the massive PR machinery of larger tech companies, these practical innovations often struggle to reach the minds of the average consumer, even when they're solving real problems we face every day.

The Laundry Robot-to-Roomba Pipeline

But it’s not like you’ll be buying this anytime soon anyway, for a few reasons:

• It's incredibly challenging from a technical perspective. While language models can train on existing text data, physical automation requires real-world data, often created from scratch.

• It's extremely expensive to develop. Startups tackling these "mundane" tasks often work with resources that pale in comparison to tech giants like OpenAI (even if they are backed by OpenAI). They need specialized equipment and data that can't simply be downloaded from the internet.

• The path from prototype to consumer product is long and complex. Even successful demonstrations don't necessarily translate to reliable, affordable home devices.

These challenges aren't new to domestic automation. In fact, the story of bringing revolutionary household technologies into the average home has always been complicated—not just technically and economically, but socially as well. And unlike our modest laundry-folding robot, these previous domestic innovations came with marketing campaigns that made them impossible to ignore.

Boxed Cake Mix and Keeping a Home

I don’t know that Joanna intentionally chose laundry and dishes are her examples, but it stood out to me when I revisited that quote to write this newsletter.

It made me think of boxed cake mix. And that made me think of the automation of domestic household work during the Industrial Revolution.

(Don’t ask how my brain works, just go with it)

The 20th century brought wave after wave of supposedly revolutionary domestic technologies—each promising to liberate women from household drudgery. In practice, rather than reducing domestic workload, these technologies often reshaped and sometimes even increased it. Washing machines made the physical act of washing easier but raised expectations about cleanliness. Vacuum cleaners didn't just clean carpets—they created new standards for how clean homes should be.

For example: boxed cake mix. When first introduced in the 1930s as complete mixes (just add water), they flopped. Marketing research revealed women felt guilty using them—it wasn't "real" baking. The solution? Companies reformulated the mixes to require adding a fresh egg, creating what they called the "egg theory of marketing." This small task preserved the illusion of traditional homemaking while maintaining convenience, but it also reinforced rather than challenged domestic expectations.

Even modern domestic technologies can present a mixed blessing. Robotic vacuums promise to take the chores of sweeping and vacuuming off of our hands, yet users often have to step in — redirecting when the robot gets stuck, emptying the bin, removing items that have gotten tangled in the brushes — only to end up with partially clean floors that they’ll have to finish cleaning up manually anyway.

(also, as someone who previous owned a Roomba, mine was extremely loud, to the point that it was easier for me to go back to vacuuming/sweeping myself)

Through this lens, the novelty of an AI-driven laundry-folding robot invites questions about real-world utility. Will it truly reduce workload, or will it, like its predecessors, come with new forms of labor? Will we actually need to separate our whites from our colors? Will we need to maintain complex machinery?

And most importantly, will it be able to fold a fitted sheet?

This isn't to say we shouldn't be excited about domestic automation. Rather, we should approach it with clear eyes about both its potential and its limitations. The history of domestic technology suggests that true liberation from household tasks isn't just about the technology itself—it's about the social structures and expectations that surround it.

What Would Actually Help You?

All of this isn’t to say that we can’t be excited about laundry-folding robots, or domestic automation more broadly, only that I think it’s still important to consider the context in which these tools exist and the histories behind them. We’ve been talking about in-home robots for decades, yet none of them have come to fruition in a way that impacts us as meaningfully as we’d like. And even if we do get in-home laundry folding robots one day, it’s hard to know how much they’d really relieve the day-to-day workload, especially when we look at past automation “wins” that added complexity rather than freeing up time.

I think about this a lot as it relates to the daily firehose of new AI tools (or updates to existing ones) — each promising to make your life easier by solving a problem you may or may not actually have. As someone who tries a lot of those tools, my backlog is filled with abandonware—promising innovations that ultimately created more work than they saved. They're the digital equivalent of that fancy kitchen gadget that's supposed to save you time but ends up gathering dust in a drawer because it's just too much hassle to use and clean.

Yet, there are tools that have genuinely improved my life. For example, I use two different tools with overlapping functionalities to edit my videos - Gling and Descript. Both are transcription-based editing tools that can automatically cut out bad takes and edit for clarity. However, I’ve found that Gling works better for my long-form video editing workflow and Descript is better for my short-form editing workflow² . Neither tool is perfect on its own, but each serves a specific purpose that genuinely brings value to my workflow.

And isn't that what we're all really looking for? Not robots to replace us entirely or AI to completely automate our lives, but tools that help us reclaim time for the things we value. So while we wait for the laundry-folding robots, I'll keep digging through my laundry pile, dreaming of a future with a Roomba that folds my fitted sheets.

In Other News

• Robert Downey Jr. Refuses to Let Hollywood Create His AI Digital Replica: ‘I Intend to Sue All Future Executives’ Who Recreate My Likeness (Zack Sharf / Variety)

  • There’s an Iron Man/Doctor Doom joke here somewhere. Also, RDJ is the lead in a Broadway play critiquing AI execs right now!

• The case for targeted regulation (Anthropic)

  • Disclaimer: I know Jack Clark, who runs their policy arm, but I do appreciate the active approach that Anthropic is generally taking to AI regulation advocacy.

• Should you feel guilty about using AI? (Adam Clark Estes / Vox)

  • This is actually an exploration of climate impact, not all the other reasons you might feel guilty 😅

• This Is a Glimpse of the Future of AI Robots (Will Knight / WIRED)

  • Physical Intelligence did get covered in WIRED if you want to read more!

• AI Platforms Are Offering Up George Floyd Simulations (Brian Contreras / Inc.)

  • Remember Character AI from last week’s newsletter (and the sad news that came out right after)? Yeah.

The Last Loop

speaking of mundane problems that AI might be able to fix…

(see also: all the text messages)

1  They are financially backed by OpenAI, Sequoia Capital, Thrive Capital, Lux Capital, and Khosla Ventures.

2  tl;dr - the level of editing that I typically need to do for long-form videos requires Premiere Pro/Resolve/FCP because I’m doing layered timeline edits. On the other hand, for short-form video, all I really need most of the time is to cut down the video and add captions. Descript lets me do that in one place very easily, but as soon as you need to add layers to a timeline (green screen, graphics) it becomes much more of a pain to use.