Hands on with Stickerbox, the AI-powered sticker maker for kids


There’s a new AI-powered toy for kids called Stickerbox, and, before you groan, I’m here to report that it’s surprisingly fun.

Stickerbox, a product born out of Brooklyn-based startup Hapiko, is a voice-activated sticker printer. The device takes whatever creative idea you have in your head and transforms it into a printed sticker that you can then color, peel, and stick anywhere.

Before trying the device itself, I have to admit I came with a preconceived negative bias — as did my fellow tester (my daughter). Our initial reactions were similiar: “An AI that prints stickers? I’d rather design and print my own.”

After trying the review unit sent by the company, we were won over.

Stickerbox, I realized, could represent a new form of creative play — and one that doesn’t outsource the child’s imagination to an AI model as much as you’d think.

Image Credits:TechCrunch

Testing the AI sticker printer

The $99.99 toy itself is a small, bright red box with a black-and-white screen and a big, white “push-to-talk” button on top. It ships with three rolls of paper, which equates to 180 stickers, as well as a power cord and colored pencils.

The box’s color scheme is reminiscent of the Etch A Sketch, which makes sense, given that the Stickerbox feels like a modern spin on that concept. In the Etch A Sketch’s case, you have to learn how to control different knobs to create the image in your mind. With Stickerbox, those “knobs” are replaced with something more abstract: the voice commands you use to prompt the AI model.

Kids aren’t thinking about how to be better prompt engineers, of course; they’re just exploring their imagination and having fun seeing their ideas come to life. Any improvement in their prompting abilities is a side effect.

To initially set up the device, a parent will need to help. Much like adding a smart speaker to your home’s Wi-Fi, you have to first connect to the Stickerbox’s Wi-Fi, then enter the information to connect with your home network. The setup process, which only took a minute, went off without a hitch.

Image Credits:Stickerbox

Using Stickerbox is simple. You push the button, describe an image out loud, then release the button to see your text appear on screen, followed by an AI-generated image as the printer spits out a physical copy.

There’s a serendipity to an experience in which you’re thinking of an idea and then holding it in your hand in a matter of seconds.

The device’s thermal image printer requires no ink, and the paper is BPS and BPA-free, making it safe to use.

The printed sticker is easy to tear off and then can be colored in with the colored pencils that come with the device. Your own crayons and markers also work. This combines the somewhat dopamine-driven experience of thinking up new things to print with the more calming or meditative aspects that come with coloring, similar to giving kids a coloring book.

This ended up offering a healthy balance between using potentially addictive tech and then slowing down to engage in a real-world activity. It also helped to address potential boredom.

Image Credits:TechCrunch

The more you use the Sitckerbox, the more you realize how complex your prompts can be. You don’t just have to ask for a basic image, like a “magical unicorn,” you can speak to the Stickerbox with long, train-of-thought commands, and the AI parses what you mean. (This is particularly useful given that kids don’t tend to explain things in a straightforward fashion.)

Making “AI for Kids”

Hapiko, the company behind Stickerbox, was founded this year by CEO Arun Gupta and CTO Robert (Bob)Whitney. The pair originally met while working at the e-commerce marketplace Grailed, where Whitney was director of engineering and Gupta was CEO. (The company sold to GOAT Group in 2022.)

Before Grailed, Gupta founded and launched the Y Combinator-backed hardware sleep tracker WakeMate.

Stickerbox co-founders Arun Gupta (CEO) and Bob Whitney (CTO)Image Credits:Stickerbox

Whitney, meanwhile, had worked as the director of engineering at The New York Times’ Games division, as the publisher pivoted from offering just crosswords to becoming a full-fledged gaming app, acquiring Wordle and launching other games like Connections and Strads. While that experience taught him a lot about what makes a great consumer-facing product, his later stint at Anthropic gave him a first-hand look at the advances in AI technology.

However, it was his experience as a father that inspired Stickerbox.

When his son asked for a coloring page he didn’t have on hand, he turned to ChatGPT to make a printable image.

“I made it for him — a tiger eating ice cream. And he had never seen a printer before. I got out from under the bed, our HP brother printer — literally dusted it off and printed it for him, and he ran off happily and started coloring it,” explained Whitney. “But, a minute later, the gears were turning, and he came back to me, and he was like, ‘I want a lizard riding a skateboard.’ And I was like, okay, cool, let me make that for you.”

His son was so thrilled with the process of being able to say something and see it come to life that he realized there could be something to this.

“I just saw this look on his face of magic — like pure magic,” noted Whitney.

The co-founders were also thinking about how AI technology offered so many novel experiences, but most weren’t made for kids.

“Nobody’s building AI specifically for kids. So that’s what we’re looking for,” said Gupta. “What are the right guardrails? What are the right ways? What are the right products?”

They realized that kids have great imaginations, ideal for working with an AI image model.

“[They have] endless imagination and creativity … they’re learning new things every day. Every week, they’ve got a new obsession. We’re literally the first people in the world, I think, to put an image model inside of a box,” Gupta said.

Built for Updates

Under the hood, Stickerbox actually uses a combination of AI models, including its own proprietary tech focused on making the device kid-safe. It won’t respond to requests for harmful content, like violence or sexual imagery, and it filters out swear words. And if you try a somewhat more innocuous command, like “boobs,” it just prints a random sticker that may be vaguely related to the word. (For instance, you might get a generic cartoon girl, but not a large-chested one.)

After trying and failing to get a naughty result, most kids will likely go back to just prompting the device for silly images instead.

“We want to be the trusted brand for parents where you don’t have to look over your kid’s shoulder and be like, ‘what are they doing? How are they using this?,” said Gupta.

For now, the company generates some revenue from the device sales, but it keeps the cost of restocking paper low. It’s just $5.99 for three rolls, which equates to 180 stickers. (It’s currently running a promotion that offers six rolls with every purchase now.)

Over time, the team plans to explore adding premium features, including a way to upload your own image to imagine yourself in fantastic scenarios or collaboration tools.

As a Wi-Fi-connected device, the Stickerbox is regularly updated with new firmware and features. In tests, for instance, we were able to print some recognizable characters, but a more recent update added new guardrails to guide kids toward more original designs.

A soon-to-launch companion app also lets you view past creations and save favorites, and could ultimately serve as the home for premium features.

Stickerbox is backed by $7 million in funding from Maveron, Serena Williams’ Serena Ventures, the Allen Institute’s AI2 incubator, and various angels, including Matt Brezina, and product leaders from other consumer apps.



The man betting everything on AI and Bill Belichick


Lee Roberts meets me at the University Club of San Francisco on a Friday morning, hours before his football team will lose to Cal in heartbreaking fashion – a fumble at the goal line, because little about the University of North Carolina at Chapel Hill’s expensive experiment with Bill Belichick has gone according to script.

But Roberts, the Chancellor of UNC, doesn’t know this yet. Right now, he’s in California to talk about artificial intelligence, which is both forward thinking and also – I’d hazard a guess – a welcome distraction from a lot else happening at the 235-year-old school.

“No one’s going to say to [students after they graduate from college], ‘Do the best job you can, but if you use AI, you’ll be in trouble,’” Roberts tells me, leaning into his central thesis about preparing students for the real world. “Yet we have some faculty members who are effectively saying that to students right now.”

Roberts has joined me in between other meetings in the city with AI companies because UNC has decided to make AI its north star. It’s a business bet, really. Roberts spent three decades in finance, most recently as managing partner of a private investment firm, and served as state budget director under a Republican governor. He taught budgeting as an adjunct at Duke but never worked in academic administration before becoming UNC’s interim chancellor in January of last year, a post made permanent eight months later.

Never mind that the university just lost 118 federal grants totaling $38 million as part of a sweeping effort by the federal government to terminate more than 4,000 grants across 600 institutions. Never mind that more than 900 people last year signed a statement saying they wouldn’t recognize Roberts as chancellor when he was appointed, calling the process a political “coronation” rather than a search. Never mind that Belichick’s much-touted return to football is currently a 2-4 trainwreck, with write-ups about the team’s dysfunction becoming routine fodder for sports writers. Roberts is focused on the future.

At UNC, Roberts explains, there’s a spectrum between faculty who are “leaning forward” with AI and those who have “their heads in the sand.” It’s diplomatic phrasing for what is clearly a culture war playing out in faculty lounges across UNC and – it’s probably safe to assume – at other schools across the world. While one UNC professor is assigning more research than students could complete without AI (“much closer to a real world scenario,” says Roberts), others are treating chatbots like anabolic steroids. If you use them, you’re cheating.

“We have 4,000 faculty members,” Roberts says, as a cable car clatters past the open window beside our table. “And they pride themselves, as they should, on their independence and autonomy in how they teach their classes.”

Techcrunch event

San Francisco
|
October 27-29, 2025

It sounds a little like code for: tenured professors can’t be forced to do anything. So Roberts is creating “incentive-based programs” to move the ball forward, like promoting one of the school’s deans into the role of Vice Provost for AI at the university. That individual, Jeffrey Bardzell, has been a professor for more than 20 years and has “experience both in technology and as a humanist,” says Roberts, adding that Bardzell is “exceptionally well-placed to help the faculty as a whole come further up to speed.”

UNC is barreling ahead on other fronts in the meantime. In its biggest development to date, the university announced this month that it is merging two schools – the School of Data Science and Society and the School of Information and Library Science – into one yet-to-be-named entity with AI studies at the center of the Venn diagram.

UNC isn’t alone in betting big on AI. At least 14 colleges now offer bachelor’s degrees in AI, and universities like Arizona State University have made headlines for integrating AI tools across all disciplines.

Still, creating this new school has worried some of the school’s library science students, who wonder what will happen to their degrees, according to a report in The Daily Tar Heel, the school’s independent student newspaper. At least one faculty member also complained anonymously in a statement to the paper, saying Roberts pushed for the school without a “cogent idea” of what it will entail, adding that the “careers of faculty, staff and students at both of these schools are being sacrificed to Roberts’ ego.”

Roberts tells me the implementation will be collaborative, not top-down. He’s also clear that the move is proactive, not reactive. “This is not about shutting anything down,” he says. “It’s not predominantly a cost-savings move,” he continues, a possible nod to those lost federal research dollars, which amount to 3.5% of UNC’s overall research funding.

Roberts doesn’t minimize the devastation of losing grants – “in many cases, [people] lose their life’s work,” he acknowledges – but he’s also quick to note that 3.5% is “well within our average annual variance.” He adds that he has been spending “a lot of time talking with policymakers and legislators in Washington about the tremendous good that federal research funding represents. We need to be especially vigilant right now, when there’s so much uncertainty around [these dollars] that it’s really changing the basic structure of how large research universities have been funded.”

Of course, it raises questions about resources in the aggregate. Though UNC’s AI push is the topic du jour, I ask about the $10 million the school is paying Bill Belichick annually as part of a five-year deal signed back in January. I’m from Cleveland, I tell Roberts. I remember when Belichick cut Browns quarterback Bernie Kosar, a hometown hero. The city never forgave him.

Roberts is ready for this. College sports are changing rapidly, he says. Every peer institution spends at least as much on football; many spend more. Football drives revenue for 28 other sports. UNC just won its fourth national championship in women’s lacrosse, its 23rd in women’s soccer. None of that happens without football money.

“If we had hired somebody else and we were [down some games], everybody would be saying, ‘Hey, man, you could have had Bill Belichick,’” Roberts offers.

In reality, the prevailing narrative about Belichick isn’t just about wins and losses. Even if, ultimately, that’s exactly what it’s about, numerous outlets have published stories describing chaos inside the program, with players, parents, coaches, and administrators all painting a picture of a legendary NFL coach whose style doesn’t translate to college kids.

But Roberts isn’t making decisions based on “a couple of news stories,” he says. “Coach Belichick, in my view, has done a really good job integrating with our campus,” Roberts says. He shows up at other teams’ games. He sends pizzas to fraternities on Saturday nights. He grew up on a college campus – his father was the coach at Navy.”

Hours after our conversation, UNC will lose to Cal when wide receiver Nathan Leacock loses control of the ball just as he’s crossing into the end zone for what would have been the game-winning touchdown. I can only imagine what the immediate reaction is like back in Chapel Hill.

My sense is that Roberts will brush it off. He may never be forgiven for not having a traditional academic background, but he also can’t afford to care that this bothers some people. I note that the 900-person petition took issue with the fact that, among the top 50 universities, Roberts is the only leader without higher education administration experience. The petition ran in The Daily Tar Heel, which has been critical of Roberts’ chancellorship throughout.

“I don’t think it was 900 students,” Roberts corrects me. “I think it was 900 people, regardless of whether they were students, faculty, staff, or just people in the world who signed an online petition.”

I ask how he felt about the whole episode. “No matter what your background was before you came into a job like this, you would have a lot to learn,” Roberts says. If you were a provost, you’d know nothing about “the business or finance or budgetary or political or operational or real estate sides of the university.” If you came from business, you’d need to learn the academic side.

It’s a reasonable point. The modern university chancellor is part CEO, part diplomat, part fundraiser, part sports executive. Presumably, no one arrives with all the skills required. “I think almost no matter what you did previously before coming into a job like this, there would be a learning curve,” Roberts says.

What strikes me about Roberts is that he seems relatively unbothered. The federal funding cuts are within the normal range. The Belichick hire is a wait-and-see situation. As for some of the faculty’s resistance to AI, it’s a puzzle to be solved.

He’s also making big bets just as higher education is being squeezed every which way. Federal funding is uncertain. Birth rate declines threaten future enrollment. The value of a college degree is in question, with more students graduating to find the only jobs available to them are low-wage gigs they could have landed without spending staggering amounts on college. Now AI threatens to upend the whole model.

But Roberts sees opportunity where others might see a crisis. He also thinks the window of opportunity is shorter than some might imagine. “The challenge of AI is that we have to work relatively quickly, and we also have to cooperate across academic disciplines,” he says. “And those are two things that universities, historically, are not especially good at.”

Whether Roberts’ game plan works remains to be seen. What’s clear is that he’s betting moving fast and shaking things up is better than moving slowly and preserving tradition at highly ranked UNC.

“We’re going to try to make Carolina the number one public university in America,” he tells me.

It’s an ambitious vision, and as he delivers it, for better or worse, he sounds very much like a Silicon Valley CEO.

To hear this interview with Roberts, listen to TechCrunch’s StrictlyVC Download podcast; new episodes drop every Tuesday.

OpenAI’s experimental model achieved gold at the International Math Olympiad


OpenAI has achieved “gold medal-level performance” at the International Math Olympiad, notching another important milestone for AI’s fast-paced growth. Alexander Wei, a research scientist at OpenAI working on LLMs and reasoning, posted on X that an experimental research model delivered on this “longstanding grand challenge in AI.”

According to Wei, an unreleased model from OpenAI was able to solve five out of six problems at one of the world’s longest-standing and prestigious math competitions, earning 35 out of 42 points total. The International Math Olympiad (IMO) sees countries send up to six students to solve extremely difficult algebra and pre-calculus problems. These exercises are seemingly simple but usually require some creativity to score the highest marks on each problem. For this year’s competition, only 67 of the 630 total contestants received gold medals, or roughly 10 percent.

AI is often tasked with tackling complex datasets and repetitive actions, but it usually falls short when it comes to solving problems that require more creativity or complex decision-making. However, with the latest IMO competition, OpenAI says its model was able to handle complicated math problems with human-like reasoning.

“By doing so, we’ve obtained a model that can craft intricate, watertight arguments at the level of human mathematicians,” Wei wrote on X. Wei and Sam Altman, CEO of OpenAI, both added that the company doesn’t expect to release anything with this level of math capability for several months. That means the upcoming GPT-5 will likely be an improvement from its predecessor, but it won’t feature that same impressive capability to compete in the IMO.

This AI-powered startup studio plans to launch 100,000 companies a year — really


Henrik Werdelin has spent the last 15 years helping entrepreneurs build big brands like Barkbox through his startup studio Prehype. Now, with his new, New York-based venture Audos, he’s betting that AI can help him scale that process from “tens” of startups a year to “hundreds of thousands” of aspiring business owners.

The timing certainly feels right. Mass layoffs across a variety of industries have left many workers reconsidering their career paths, while AI tools have markedly lowered the barrier to building digital products and services. At the center of that Venn diagram is Werdelin’s latest venture, with its promise to help “everyday entrepreneurs create million dollar AI companies” without requiring technical skills.

Werdelin’s journey from Prehype to Audos reflects the broader transformation happening in entrepreneurship right now. At Prehype, the focus was on working with tech founders to build traditional startups, the kind that might raise millions and aim for billion-dollar exits.

Now, he tells TechCrunch, “What we’re trying to do is take all that knowledge, all the methodology that we’ve created over the years of building all these big companies, and really trying to democratize it.”

The idea is that “everyday entrepreneurs” may sense a shift is afoot but may not be keen to experiment with so-called AI agents or know how to reach customers. Audos is more than happy to help them, supplying these individuals with AI tools to build sophisticated products using natural language, and taking advantage of social media algorithms to find them their niche customers.

“Facebook and a lot of these platforms, they are just incredible algorithms, and they’re incredible at figuring out [how to reach your customer] if you define a customer group,” says Werdelin, who co-founded Audos with his Prehype partner Nicholas Thorne. In fact, Audos uses this system to quickly test whether a founder’s business idea has sustainable customer acquisition costs.

The approach seems to be working. Audos has helped launch “low hundreds” of businesses since its beta launch, with its own customers discovering the platform through Instagram ads asking “Have you ever thought about starting something, but don’t know where to go?” Among them, Werdelin says, are a car mechanic who wants to help people evaluate repair quotes, an individual who is selling “after death logistics” services, virtual golf swing coaches, and AI nutritionists. In a winking reference to billion-dollar businesses, or so-called unicorns, he calls these one- and two-person teams “donkeycorns.”

All went through the same process: they clicked on Audos’s ad, its AI agent launched a conversation to figure out the problems these individuals want to tackle and who they want to serve, and, when it was satisfied with the answers, Audos got them in front of potential customers as fast as possible.

As for returns, Audos operates on a fundamentally different model than traditional accelerators or venture capital. Instead of taking equity, the company takes a 15% revenue share from the businesses it helps launch. In return, founders get up to $25,000 in funding, access to those AI-powered business development tools, and help with distribution, primarily through paid social media advertising.

“We’re not taking any equity in their business,” Werdelin says, partly because “we don’t think these companies might ever get sold. What we’re really inspired by are the mom-and-pop shops that are the backbone of our society.”

The revenue share continues indefinitely, similar to platform fees charged by Apple’s App Store. For founders, that means giving up a significant portion of their revenue in perpetuity — a 15% cut that could cost entrepreneurs hundreds of thousands of dollars over time. Some will undoubtedly see that trade-off as worthwhile; others might question whether the long-term costs justify the benefits.

Audos’s value proposition raises other questions given how quickly the landscape is changing. While Werdelin emphasizes helping founders build relationships with customers, it’s unclear how much of that work the AI agents can actually handle. There’s also the matter of differentiation. As Werdelin readily acknowledges, “the world is full of these tools” and they’re getting better rapidly. What happens when entrepreneurs can access similar AI capabilities without paying a permanent revenue tax?

Audos’s VCs don’t sound worried about those scenarios. True Ventures led Audos’s $11.5 million seed round, with partner Tony Conrad explaining the appeal in a Zoom call this week. In addition to having confidence in Werdelin and Thorne, says Conrad, “I think there are just lots and lots of people” who might eagerly embrace the opportunity to work with a platform like Audos.

Conrad draws parallels to Instagram’s $1 billion exit with just 13 employees, suggesting that AI could enable even more leverage, even if Audos — which itself employs just five people altogether currently —  isn’t chasing unicorns. As Werdelin explains it, “What we’re after here is the millions of people who can create million-dollar businesses or half-million dollar businesses that are real and life changing.” 

Adds Werdelin separately of why he spun up Audos, “What we’re trying to do is to figure out how you make a million companies that do a million dollars turnover. That’s a trillion dollar turnover business.”

It doesn’t sound crazy. Extending the benefits of entrepreneurship to people who traditionally haven’t had access to startup capital or technical skills is an increasingly compelling proposition as traditional employment begins to feel less and less stable. “We believe that there should be somebody who goes out and really helps these smaller entrepreneurs that are building something that is not venture backable,” says Werdelin. “We believe that the world is better with more entrepreneurship.”

Audos’s other investors include Offline Venture and Bungalow Capital, along with numerous high-profile angel investors – Niklas Zennstrom and Mario Schlosser among them.

Pictured above, left to right, Audos co-founders Nicholas Thorne and Henrik Werdelin.

Leak reveals Grok might soon edit your spreadsheets


Leaked code suggests xAI is developing an advanced file editor for Grok with spreadsheet support, signaling the company’s push to compete with OpenAI, Google, and Microsoft by embedding AI copilots into productivity tools. 

“You can talk to Grok and ask it to assist you at the same time you’re editing the files!” writes reverse engineer Nima Owji, who leaked the finding. 

TechCrunch has reached out to xAI to confirm the findings and learn more. 

xAI hasn’t explicitly detailed its strategy for pursuing interactive, multimodal AI workspaces, but it has dropped a series of announcements that point to how the company is thinking about these tools. In April 2025, xAI launched Grok Studio, a split-screen workspace that lets users collaborate with Grok on generating documents, code, reports, and browser games. It also launched the ability to create Workspaces that let you organize files and conversations in a single place. 

While OpenAI and Microsoft have similar tools, Google’s Gemini Workspace for Sheets, Docs, and Gmail appears to be the most similar to what xAI is reportedly building. Google’s tools can edit Docs and Sheets and allow you to chat with Gemini while looking at or editing documents. The difference is that Gemini Workspace only works within Google’s own ecosystem. 

It’s not clear what types of files xAI’s editor might support aside from spreadsheets, or whether xAI plans to build a full productivity suite that could compete with Google Workspace or Microsoft 365.  

If Owji’s findings are true, the advanced editor would be a step towards Elon Musk’s ambitions to turn X into an “everything app” that includes docs, chat, payments, and social media.



Xbox + AMD: Powering the Next Generation of Xbox


Xbox + AMD Hero Image

Xbox + AMD: Powering the Next Generation of Xbox

.post-template-default .xwsrc-block-content-block .wp-block-column.flex-basis-50.push–25.column–content {
flex: 100%;
margin-left: 0;
}

The post Xbox + AMD: Powering the Next Generation of Xbox appeared first on Xbox Wire.

Google tests Audio Overviews for Search queries


Google Search is experimenting with Audio Overviews for certain Search queries, the company announced on Friday. The feature was first introduced to NotebookLM, Google’s AI-based note-taking and research assistant.

The tech giant says Audio Overviews will use its latest Gemini models to give users another way to absorb and understand information.

“An audio overview can help you get a lay of the land, offering a convenient, hands-free way to absorb information whether you’re multitasking or simply prefer an audio experience,” Google explained in a blog post.

The feature is available starting today in Labs, Google’s experimental program. The company says users will see the option to generate a short Audio Overview if Google thinks it would be useful based on their specific query. Once you generate an Audio Overview, you will see a simple audio player with play/pause controls, a volume button, and the option to adjust the playback speed.

Image Credits:Google

Google will display links in the audio player to show where it’s getting the information from. If you want to learn more about a topic after listening to an Audio Overview, you can click on the links in order to dive deeper into your search.

You can give a thumbs up or thumbs down on each Audio Overview, and the experiment as a whole in Labs.

In NotebookLM, Audio Overviews give users the ability to generate a podcast with AI virtual hosts based on documents they have shared, such as course readings or legal briefs. Google also brought Audio Overviews to Gemini in March.

Audio Overviews in Search builds on AI Overviews, the AI-generated summaries Google supplies for certain Google Search queries. With Audio Overviews, Google is targeting people who are auditory learners or want more accessible ways to get information.

It’s worth noting that today’s announcement comes a few days after a Wall Street Journal report found that Google’s AI Overviews and other AI-powered tools are killing traffic for news publishers.

Anthropic brings web search to free Claude users


Anthropic is continuing to trickle down features to its free users. The latest one to make the leap out of subscriber-only mode is web search, which the company to its AI chatbot Claude in March. According to Anthropic, connecting Claude to the web allows it to deliver more accurate responses based on the most up-to-date information available online. This feature is available to all Claude users starting today.

In addition, Anthropic has begun beta tests for voice mode on its mobile apps. This option lets users interact with Claude in natural conversations in an expansion of the platform’s existing dictation tools. There will be five voice options available to assign to Claude, and the AI assistant can provide full transcripts and voice mode summaries after a conversation.

May has been a busy month for Anthropic, which just launched last week. Opus 4 is a powerful coding-focused system that can use multiple tools in parallel and can run for several hours at a time, while Sonnet 4 is a hybrid reasoning model designed to move between quick queries and more complex ones. The current beta testing of voice mode will default to Sonnet 4.

Macs finally get a taste of an overhauled Mail app


Apple redesigned the Mail app on iPhones with the release of iOS 18.2 update back in December, but strangely skipped the treatment for iPads and Macs. The company has finally made a course correction with the macOS 15.4 and iPad OS 18.4 developer beta updates, which are now available for testers.

The biggest change introduced by the new Mail app are categories. All your emails are now neatly slotted across four categories. Here’s a brief breakdown of how it works:

  1. Primary: For personal messages and time-sensitive content.
  2. Transactions: A section for keeping a tab on confirmation emails, receipts, and shipping-related alerts.
  3. Updates: All the content that you’ve signed up to receive via an email agreement, such as news, newsletters, and social media updates.
  4. Promotions: This is the section where you get marketing and shopping material, such as coupons and sales-related notifications.

Apple, however, notes that if any of your emails across the last three categories contain time-sensitive details, they will appear in the Primary bracket. Apple is also taking an approach similar to social media profiles, when you open messages from a specific sender within these three categories.

Explanation of categories in Apple Mail app.
Nadeem Sarwar / Digital Trends

“When you tap a Transactions, Updates, or Promotions message, a digest view of messages from that sender opens,” says the company. This is a great way to catch up on activities such as alerts from your banking service provider, as you see all your payments neatly slotted in a vertically-scrolling card-like format.

Of course, you can always choose to disable categories and enable the classic list view that shows all your emails in the same order as they landed in your inbox. The idea is neat, as it separates unimportant jargon from relevant communication, but it’s not perfect.

The arrival of the updated Mail experience on iPads and Macs solves a big problem, even though it add some versatility, like letting users create their own categories. iPhones have had the new interface for a while now, and if you got used to it, not having a consistent experience on your Mac or iPad was a bummer.

With the arrival of macOS 15.4 and iPadOS 18.4’s first developer beta updates, there is finally some respite. It’s now only a matter of time before the developer and public beta testing comes to an end, and the overhauled Mail experience is available across the entire mobile and desktop ecosystem.

As far as other AI-powered tricks in the Mail app are concerned, Mac users can also take advantage of email summaries, smart replies, and priority messages.






Perplexity has its own ‘Deep Research’ tool now too


In a on Friday, Perplexity introduced a new tool called Deep Research that it says can conduct “in-depth research and analysis” to deliver detailed reports in response to your questions, and it’s free for limited use. It comes just a couple of weeks after OpenAI announced its own users… which itself followed Google’s December announcement of . Perplexity’s tool is available only on the web to start, but it will hit the iOS, Android and Mac apps soon too.

To view this content, you’ll need to update your privacy settings. Please click here and view the “Content and social-media partners” setting to do so.

Perplexity says its Deep Research “excels at a range of expert-level tasks — from finance and marketing to product research” and takes about 2-4 minutes to come up with an answer, during which it “performs dozens of searches, reads hundreds of sources, and reasons through the material.” Once finished, its reports can be shared or exported as a PDF. The company claims it outperforms competitors — like OpenAI’s o3-mini and o1, and DeepSeek-R1 — on the Humanity’s Last Exam benchmark, earning a 21.1 percent accuracy score (though this is lower than OpenAI’s Deep Research scored).

Free users will be limited to five queries per day, while Pro subscribers will get 500, according to a tweet from the company.