最新 AI 资讯
NEA’s Tiffany Luck says enterprises are still figuring out their AI ROI
Loading the player… Tokenmaxxingwas the hottest trend in Silicon Valley earlier this year, with CEOs encouraging employees to push AI usage as far as it would go.Then the bill came due. Uber reportedly blew through its annual AI budget in a few months, some companies cut Claude licenses for parts of their org, and Meta killed its internal leaderboard. This tension between hype and ROI is exactly whereNEA partner Tiffany Lucklives these days. She got her start convincing companies that e-commerce was the future, and now she’s all in on AI, especially when it comes to the possibilities for “magic moments” in the consumer business. On this episode of TechCrunch’sEquitypodcast, Luck joins Rebecca Bellan to talk about the future of personal agents, her thoughts on this year’s AI IPOs, and how startups are stepping in to help enterprises track return on AI spend. Subscribe to Equity onYouTube,Apple Podcasts,Overcast,Spotifyand all the casts. You also can follow Equity onXandThreads, at @EquityPod.
View
After unveiling ridiculously expensive AR glasses, Snap’s stock takes a dive
Snap’slong-awaited AR glasses, Specs, didn’t have the best debut. The company’s stockhasn’t been on the healthiest trajectorylately. It’s dropped 30% over the past year. Following Specs’ launch, it sank more than 5% — falling from $5.86 a share on Tuesday to a low of $4.83 on Wednesday morning. As of this writing, the stock still hasn’t recovered the position it held prior to the announcement. The big concern surrounding Snap’s new smart glasses — which the company has been working on for over a decade — is the cost: The company maintains they will retail at nearly $2,200 apiece. It’s worthy of note that Snap’s core user demographic — teenagers — are not typically equipped with that kind of pocket change, leading onlookers to question the profitability path for the new product. Snap’s CEO, Evan Spiegel, didan interviewwith CNBC on Tuesday (during which he sported the new glasses) and, when questioned about the hefty price, responded: “The most important way to think of Specs is as a computer, and so they’re comparably priced to other high-end computers or high-end laptops.” Spiegel further justified the cost by saying that Specs occupies a unique space in the AR market between glasses like Meta’s Ray-Bans — which cost a lot less but provide significantly less compute power — and bulkier headsets like the Apple Vision Pro, which are powerful but very expensive. Spiegel said his product was both “highly wearable but also incredibly capable for immersive computing.”
View
Roelof Botha joins SpaceX’s board of directors
Former Sequoia Capital managing partner Roelof Botha is joining SpaceX’s board of directors, less than a week after the company went public in the largest IPO ever. SpaceXannouncedthe appointment in a filing with the Securities and Exchange Commission on Wednesday. It said Botha was appointed “to fill the existing vacancy on the Board” and that he will serve until SpaceX’s next annual shareholder meeting. He will also join the SpaceX board’s audit committee. Botha didn’t immediately respond to a request for comment. Botha “brings extensive public company experience along with a deep audit committee background, having served on the boards and audit committees of numerous public companies,” SpaceX wrote in the filing. He stepped down fromhis role as Sequoia’s leader late last yearas the firm was dealing with blowback against partner Shaun Maguire, who had made comments attacking then-New York City mayoral candidate Zohran Mamdani. The SpaceX filing also discloses that a “family member” of Botha’s has worked at the company “since January 2025 as a member of the enterprise operations team.” SpaceX says this family member’s “compensation exceeded the $120,000 reporting threshold” but noted the compensation is “generally commensurate with their peers’.” Botha is on many public company boards, but none quite like SpaceX, where CEO Elon Musk enjoys near-absolute control and shareholders have little power. Musk has more than 80% of the now-public company’s voting power, and owners of SpaceX stock will haveseverely limited opportunitiesto challenge him should they disagree with his actions. Musk also has control over all changes to the board’s makeup,filings show. Botha has experience with Musk, though. Musk brought Botha — who is also from South Africa — in to run the finance division of PayPal in 2000. Botha started at the payments company in March of that year, according to his LinkedIn profile. Musk was pushed out as CEO of PayPal in September 2000. “I’ve known Elon for over 25 years,” Botha toldFortunelast year in an interview during Musk’s time running DOGE. “He was the first person to offer me a job in America. He believed in me when I was an unknown student at Stanford. I have a lot of appreciation and understanding that he’s not perfect. None of us are. He deeply cares about doing the right thing.” The addition of Botha brings SpaceX’s board to nine directors. He joins Musk confidants Ira Ehrenpreis, Antonio Gracias, Steve Jurvetson, and Luke Nosek; SpaceX chief operating officer Gwynne Shotwell; Google executive Donald Harrison; and VC Randy Glein. Musk is chairman of the board. Botha has been with Sequoia for more than 20 years, and the firm invested in SpaceX in 2019. It reportedlyowned1.5% of SpaceX heading into the IPO, giving it a position worth more than $20 billion.
View
Collecting robot training data is dirty, unglamorous work. Some AI labs are already paying XDOF to do it.
Two weeks ago, OpenAIsaidit would relaunch the robotics program it shuttered in 2021 — the latest signal that the biggest AI labs are racing to teach machines to operate in the physical world. But building capable robots requires something the AI industry doesn’t yet have, which is the training data to match that used for language models. That gap is creating a new kind of infrastructure business. Unlike LLMs that were trained on a vast sea of publicly available text, robots need data that captures physical interaction, and that kind of data barely exists. YouTube videos and footage captured by gig workers are low-fidelity and hard to reconcile with the physical world. XDOF(pronounced “ecks-doff”), emerging from stealth today, is betting that the next great bottleneck in AI isn’t models or chips, but the data feedback loop needed to teach robots how to interact with the physical world. The startup aims to build the data pipelines, collection tools, and annotation systems that frontier labs and robotics companies can’t easily build themselves — and has raised $70 million from Thrive Capital, Spark Capital, a16z, Lux, and WndrCo to do it. Co-founder and CEO Philipp Wu says XDOF, which has about 60 employees, is already working with 20 customers, including several frontier AI labs, but cannot name them. “All of the top labs are trying to pursue robotics,” Wu said. “We’ve already seen some of the downfalls of falling a little bit behind in the language model race … you don’t want to be in this type of situation where you pursue this technology too late, and everyone is in this boat where physical AI is the next frontier.” Wu ran into this problem himself as a PhD student at UC Berkeley. His focus was on enabling robots to learn skills from large-scale datasets. There was just one problem. “We didn’t have large-scale data to work with,” he told TechCrunch. “There was this chicken-and-egg problem — we first needed to actually collect data before we could even ask how to train a foundation model for robotics.” Wu and his future XDOF co-founder and CTO, Fred Shentu, worked on a project called GELLO, a low-cost teleoperation system that lets a human operator control a robotic arm to generate training data. “It ended up becoming a very influential paper in robotics, because a lot of people had similar needs and bottlenecks, and many started leveraging this type of device for data collection,” Wu said. Spotting the opportunity, Wu, Shentu, and third co-founder and Chief Operating Officer Nemo Jin launched XDOF in October 2024 to provide a data ecosystem for companies pursuing robotics models. Mindful that data provision alone can be a dead-end business, the company is also focused on data cleaning, tooling, and annotation — creating a self-reinforcing feedback loop for robot trainers. As a starting point, the company is partnering with UC Berkeley’s AI Research lab to release what it believes is the largest collection of high-quality robot training data ever assembled, dubbedABC. It includes 130,000 trajectories of robot manipulation data, 300 hours of simulation, and 100 hours of evaluations. That kind of scaled-up pre-training data has never been available to academia before. “We’ve seen in language, image generation, and other fields, that when models and data are released, the community achieves things that you wouldn’t necessarily have expected,” David McAllister, a Berkeley PhD student who helped organize the release, told TechCrunch. The team has already used the data to train robots on benchmark tasks like folding T-shirts and flattening boxes, or loading AirPods into their cases. The company plans to work across three tiers of a data pyramid. The most valuable tier is teleoperation data collected on the actual robot being deployed; next comes teleoperated robots gathering more general data, as with GELLO; and finally “egocentric” data gathered by humans performing everyday tasks, for which XDOF plans to build its own wearable sensors. “Your camera choice is going to affect the quality of your data — which is going to affect how your hand-tracking algorithm performs,” Wu said. “If you don’t design the hardware well from the start, the data you collect might have very specific problems that you didn’t anticipate.” The company plans to hire and train armies of teleoperators and egocentric data operators around the world — a labor-intensive model that raises an obvious question: Why aren’t the major labs doing this data production work themselves? “You need a warehouse of hundreds of thousands of square feet with hundreds of robots,” Wu said. “You need to maintain these robots, calibrate their physical parameters, and properly train operators.” It’s a build-out that requires focus, capital, and operational scale that most AI labs would rather outsource — which is precisely the market XDOF is betting on. The name XDOF is a play on the robotics term “degrees of freedom,” which describes the number of independent motions a robot can perform. Your arm, from shoulder to wrist, hasseven degrees of freedom. Humanoid robotics company Figure AI’s latest robot has 30. The X in the company’s name captures its ambition: “Arbitrary degrees of freedom, unlimited degrees of freedom,” Wu says.
View
The slowtech revolution is here to kill your phone addiction and rescue your attention span
When Tony Fadell entered New York City’s 28th Street Subway Station, he did not expect to come face-to-face withan advertisementfor a product he designed over 20 years ago. But there it was: a five-by-four-foot poster promoting the iPod Shuffle, luring passersby with the promise of “Zero screen time.” “The first thing was, I thought, ‘Wait a second, did somebody not change the ad?’” Fadell, known as the father of the iPod, told TechCrunch. “For somebody like me who knows that thing intimately, it’s like seeing your kid’s picture.” As Fadell stood in the train station, he was surrounded by people wearing wireless Bluetooth headphones to stream music on their phones, effortlessly accessing music libraries with over 100 million songs. This technology that we take for granted makes Steve Jobs’ early iPod tagline —“one thousand songs in your pocket”— sound antiquated. The postage-stamp-sized iPod Shuffle, which relied heavily on shuffle playback and offered little control compared to today’s streaming apps, should not appeal to a modern audience. But we have become so entrenched in technology that our various devices, apps, and algorithms mediate our every experience, from grocery shopping to dating. We’ve built smartphones that can do almost anything, but we’ve also created a constant connectedness that has become more exhausting than enriching. “People are very oversaturated and overstimulated, and they really want to have a more mindful approach to what they’re doing with their tech,” Joy Howard, CMO ofBack Market, an online marketplace for refurbished tech, told TechCrunch. “There’s this fatigue that we have with the need to optimize every single aspect of our life.” Howard and her team were responsible for the iPod Shuffle ad that Fadell was so shocked to encounter. But Howard says that demand is growing for this supposedly obsolete tech — if these devices weren’t driving sales, the company wouldn’t have shelled out for a premium ad placement in a hectic New York City subway station. For younger generations who have never known a world without social media and smartphones, there’s a certain magic to wired headphones, retro gaming consoles, CDs, and digital point-and-shoot cameras. They crave experiences that aren’t trying to monopolize their attention. Old-school cameras can’t upload photos to your Instagram story, retro games don’t spam you with gambling ads, and iPods can’t automatically play music that you’re algorithmically destined to enjoy. That’s the whole point of this movement, which Howard calls “slowtech.” “The ‘fast tech’ up until now has been all about eliminating friction… [Now], people are seeing friction as a way to create boundaries for themselves,” Howard said. “It’s so stunning to me that now people are wanting to bring friction back into their lives, and see that as a feature, rather than a flaw.” Around the same time that Fadell first pitched the iPod to Steve Jobs,Austin Murrayfounded JAMDAT, one of thefirst mobile gaming companies, which quickly went public and was sold to Electronic Arts for $680 million. “When we were pitching our company back in 2000, 2001, people were laughing at us, saying, ‘Why would anyone play games on their cell phone?’” Murray told TechCrunch. Now, investors are just as incredulous when he pitches them on hisscreen-time reduction app, MOQA, which he is building to counteract the very phenomenon he helped create. “It’s watching what happened to my kids and the people around me that hurts my soul the most,” Murray said. “When everyone is doing the same thing — meaning everyone, the average screen time is like five hours probably on a phone every day — it’s not a willpower problem. It’s a product design problem.” This desire to cut back on the time we spend using our phones, computers, and TVs has become ubiquitous —about 53%of American adults say they want to reduce their screen time. “At a certain point, I realized that willpower was insufficient to not waste time on my phone,” said writerCalvin Kasulke, whose novel “Several People Are Typing” imagines workers trapped inside a Slack workspace. He now pays forOpalandFreedom, two apps designed to limit his screen time and social media use. “I don’t need to limit my time on iMessage — that’s people who I really know! But I certainly don’t want to be wasting my time doomscrolling.” “I want to be very clear… I don’t feel smug about this. It’s embarrassing to have two different apps to limit how I use this,” Kasulke said. “I don’t think screens are inherently bad. I just think the way I was using [my phone] was worse and dumb, and now it’s a little bit less dumb.” Others have given up their iPhones altogether, opting instead for flip phones,e-ink devicesthat run Android software, or minimalist touch-screen hardware like theLight Phone. “Our customers for the last 10 years are telling us how they feel more free after switching to the Light Phone,” Light co-founder Kaiwei Tang told TechCrunch. “It’s getting more and more attention, especially among young people. We have quite a lot of the community using Light Phone as 20- to 35-year-olds, which surprised us.” Murray isn’t as optimistic about the future of “dumb phones,” though. “There’s certainly a movement of people who are just kind of anti-tech and ‘get it out of our lives,’” he said. “That’s really hard though, because then you realize you can’t do things that are now assuming you have a smartphone, like banking, or going into a hotel, or [using] credit cards.” Kasulke said if Apple ever made an e-ink iPhone, he would “f–ing donate plasma to be able to afford it.” But that’s unlikely, so he’s not particularly interested in downgrading his phone. “I’m not like a, ‘I wish I could throw this thing in the toilet and go live in the woods’ kind of guy,” Kasulke said. “My phone has some utility for my personal and professional life, but it also lives in your pocket, and it is very, very easy, and in fact, designed in some ways to be addictive and to mindlessly waste time on it.” Screen time isn’t universally bad. We’re accumulating screen time when we video chat with our family, text our friends, read news articles, maintain our Duolingo streaks, or play Wordle. But for as much as tech brings us closer to one another, it also yanks us out of the present moment. “It’s clear people want the convenience of digital, but they don’t want the annoyance of being always connected,” Fadell said. “I’ve always been like, ‘We need less screens, not more of them.’ So to have an Apple Watch with everything, like, no, no, no — I don’t want more, I want less.” It’s not surprising that Fadell’s preferences are a bellwether for the market — he’s a veteran product designer, after all. American spending on fitness trackersgrew 88%year-over-year, according to market research firm Circana, which credits screenless wearables like the Oura ring and Whoop wristband as key sales drivers. Even though these devices don’t have screens, you have to use your smartphone to see your data, which would make it even harder for Oura and Whoop users to try out something like the Light Phone. But most consumers aren’t looking to make such an extreme change as pivoting to a flip phone — instead, some are embracing even more sophisticated hardware that relies on their smartphone, but cuts down their overall screen time. Mark, a $159 AI bookmark, advertises itself as a tool to help users stop pulling out their phone to take notes while they’re reading. While some readers might find the idea of an AI bookmark to be symptomatic of the same problem that pushes people toward a digital detox, Mark founder Eason Tang sees it differently. “The way we try to brand it now is this sort of analog tool, very culturally integrated with design, film, books, and literature,” Tang told TechCrunch. We raised $1M dollars to reinvent how people read. Introducing Mark II – a $159 AI bookmark. Thread belowpic.twitter.com/eL0XsyRlgC There’s something undoubtedly absurd about using an AI bookmark to mediate your relationship with your phone, yet there is a bit of truth to Tang’s pitch — when you stop reading to take notes or snap a photo of a key passage on your phone, you’re bound to encounter some other distracting notification that interrupts your reading. Though AI developments are almost synonymous with “fast tech” culture, there’s a clear allure to the promise that AI agents could simplify our lives and give us more time away from screens. “I think that this idea that people want tools to serve them and not to dominate them is very profound,” Howard said. “I think what the ‘slowtech’ movement is about is people pushing back against the constant digital fatigue, distraction, overwhelm, so if you can use AI to do that, to kind of protect yourself… That’s what people want: more control.” The ubiquity of AI turns some consumers off from the latest products, but this isn’t their sole grievance with big tech. People are also disillusioned by these companies for continually bricking perfectly good hardware just to make us buy the latest model. Back Market, for example, rehabsdiscontinued laptopsand resells them with USB keys that can install ChromeOS Flex, which turns supposedly obsolete hardware into functioning Chromebooks. “One of our developers started finding a way to hack things that had their OS sunsetted to bring it new life. And so one of the first things he hacked was a rice cooker,” Howard said. “His rice cooker didn’t have support anymore! This is actually a really cool use of AI — like, vibe coding your own app to keep your hardware alive longer.” While slowtech adherents may not all agree about AI use, the debate is secondary to the bigger problem at play: We’ve created an ecosystem where we are so dependent on smartphones and our various apps that the whims of the tech industry can control how we cook rice. In this reality, it’s no wonder that people are so eager to disconnect that they want to downgrade to an iPod Shuffle. “People just really want to take back control of their time, their lives, their attention,” Howard said. “They’re down for whatever helps them do that.”
View
Google bets on Gemini to reinvent the smart home speaker
After years of incremental updates, Google is betting that its Gemini AI can reinvent its smart speaker. On Wednesday, the company introduced its first audio device built specifically for Gemini with the $99.99 Google Home Speaker. The new Google Home device is the first stand-alone smart speaker from the tech giant since theNest Audio in September 2020. That older device arrived at a time when smart speakers were thought of largely as handy controllers for your smart home and music-playing systems. They lacked the smarts of today’s AI chatbots, as commands often had to be phrased correctly to get things to work. The Google Home Speaker is changing that, as you’ll be able to speak using natural language requests and even make multistep requests using the phrasing you’d like. For instance, you could tell the speaker to “turn off all the lights except for my bedside lamp,” or “dim the kitchen lights, play some relaxing music, and set a timer for 20 minutes.” You can also make corrections mid-sentence as you speak instead of having to try requests again, and Gemini will understand. That means you could say something like, “Turn off the coffee maker … I mean, turn it on!” and the AI will respond appropriately, Google points out. Plus, the device will ship with 10 new voices that can have two-way conversations with you about topics that aren’t limited to smart home tasks or other simple commands. You can ask more nuanced questions and dive deeper into topics you want to learn about, as you could when speaking with Gemini on your smartphone. The speaker’s microphone can also remain on briefly when using the “Continued Conversation” feature, so you can more naturally ask follow-up questions without having to say “OK, Google” again. The device looks similar to older versions, with its 3D-knit textile wrapping and rounded 3.4 x 4.2-inch design. In the U.S., the speaker comes in Jade and Berry colors in addition to the Hazel and Porcelain options available in the rest of the world. A new ring light at the bottom will indicate if the speaker is listening, thinking, or responding. But not all of the new device’s AI smarts will be free. Google will sell Google Home Premium subscription plans for $10 per month (or $100 per year) if you want to take advantage of more powerful AI features. This includes being able to have more free-flowing conversations with Gemini Live, which you kick off by saying, “Hey, Google, let’s chat.” Home Premium can also help you ask about and make sense of activity captured on your home’s Nest cameras, or offer summaries of what happened in the home while you were out. Whether those capabilities are compelling enough to justify another monthly subscription remains to be seen, particularly when many of the device’s Gemini features are available without paying. Google will try to get you used to the advanced features by offering them for free for six months before pushing you to subscribe. If successful, Google will have reinvigorated the smart speaker lineup with generative AI and will have found a way to get some customers to pay for those technological advances. The device is available forpreordernow and will ship later this month.
View
Only 16 percent of Americans think AI will have a positive impact on society, a new study shows
Despite the fact that AI increasingly dominates our economy (it’s ahot IPO summerand we’re all just along for the ride), most Americans are not particularly optimistic about the technology’s long-term impact on the country, a new study from Pew Research reveals. In fact, although a whole lot of Americans increasingly use AI in their daily lives, most of them have neutral to negative views about it, the research reveals. Only 16% of Americans think that AI’s impact on society during the next 20 years will be positive, Pew says, while around 40% say that it will have a negative impact. A vast majority of people (67%) don’t believe that the U.S. government will do anything to meaningfully regulate AI. A similarly skeptical cohort (59%) don’t trust companies to develop it safely. Young people — that is, those people under 30 — are the ones with the most negative feelings about AI. Pew says that only 14% of this cohort believe the tech will have a positive impact on society. On top of all this, a vast majority of Americans — nearly two-thirds — also think that AI’s development is occurring too quickly. Despite all of the skepticism, a whole lot of Americans also report using AI in their daily lives on an increasingly regular basis. About a quarter of Americans say they use AI chatbots on a daily basis. Those who do are typically using the chatbots for research purposes or for work, Pew says. A vast majority of people using AI are using ChatGPT. Pew writes that 44% of U.S. adults now say they use OpenAI’s chatbot, a figure that’s more than doubled since 2023. The next most popular chatbot is Gemini (24%), followed by Copilot (17%) and Meta AI (14%), with Grok (8%), Claude (6%), and Character.ai (3%) lagging behind. There is a bit of a gender divide. While chatbot use is growing for both men and women, men still use AI more and are more enthusiastic about it, while women are more skeptical, Pew says. Men are more likely to say they use AI chatbots in their daily lives (27% versus 20% for women) and while equal shares of men and women report using ChatGPT, men more commonly report usage of other brands, such as Copilot and Grok. The report also highlights how AI is changing the ways Americans consume information. Six in 10 survey respondents told Pew that they routinely read AI-generated internet summaries (indeed, on Google, they’re pretty much unavoidable). A much smaller number report using AI to get information on fitness and dieting. There are also still a whole lot of people — about half of the country — that say they donotuse AI in their daily lives. The people who do not use AI tend to be older, while those under 50 are more likely to say that they use it. Nearly 75% of Americans aged 65 or older say that they never use AI chatbots. Those people who don’t use chatbots say they don’t because they’re not interested in them, and add that they have no intention of using them in the future.
View
World model maker Odyssey nabs $1.45B valuation backed by Amazon and other big names
Odyssey, a world model AI startup founded by self-driving vehicle pioneers CEO Oliver Cameron and CTO Jeff Hawke, has raised a $310 million Series B round at a $1.45B valuation led by Natural Capital, with Amazon, AMD Ventures, GV, and others participating. World models are the next big thing in AI beyond text- and chat-based large language models. They gather data from the physical world and simulate it with accurate physics. In Odyssey’s case, it has mimicked how Google Earth gathered data; the startupsent people out with cameras strapped to their backs. (Google drives camera-equipped cars around.) That approach makes sense given the backgrounds of the founders. Cameron was the co-founder and CEO of autonomous vehicle startup Voyage, which wasacquired by GM’s Cruise, where he later became VP of product; Hawke was an engineer atbuzzy U.K. self-driving startup Wayve. Odyssey, founded in 2023, now offers a handful of world models for a variety of use cases, from video-game creation to robotics. It is perhaps best known for producing rich, interactive video from text prompts. With the backing from Amazon, the startup says AWS is now its preferred cloud provider and it will optimize its models to run on AWS’s Trainium chips, a competitor to Nvidia’s AI chips. In addition to the VCs that participated in this unicorn-crowning round, Odyssey has corralled an impressive list of angel investors as well. These include Jeff Dean, Elad Gil, Garry Tan, Guillermo Rauch, and Cruise founder Kyle Vogt. The company has now raised $337 million to date.
View
Social media’s next evolution: user-controlled algorithms
For years, social media giants controlled what users saw in their feeds. While people could follow accounts, like posts or hide content they didn’t enjoy, recommendation algorithms controlled what was shown to them. Now, social platforms are handing over some of their power by allowing users to personalize their algorithms with the help of AI. Beyond traditional “Not Interested” buttons, apps like Threads, Instagram, and TikTok have begun introducing tools that let people train their own algorithms and influence what appears in their feeds. The shift reflects an evolution in how recommendation systems work. Social media feeds are moving away from a one-size-fits-all TV channel and toward something more like a streaming service, where users can tune recommendations to their interests and have more control over what they see. For users, the advantage of customizable algorithms is a feed tailored to their interests. For social media giants, it’s a way to boost engagement by displaying content that people are most likely to consume. Here’s a look at how social media platforms are giving users more control over their recommendation algorithms. On July 16, 2026, Threads launched a new“Your Algo” featurethat builds on the platform’s“Dear Algo” tool, which debuted in February. The “Dear Algo” tool lets users influence their feed by publishing a public post, such as “Dear Algo, show me more posts about podcasts,” to indicate the types of content they want to see more or less of in their feed. With the new “Your Algo” feature, users can make those preferences privately, without having to post publicly. Users can tell Threads you want to see more or less of certain topics and choose how long the request lasts: one, three, or seven days. For example, you could ask to see more baseball content and less stressful news. In early June, Instagram launched a new tool that lets you see and control your algorithm across your feed. The tool, called “Your Algorithm,” lets you view the topics that shape your recommendations and customize them to better fit your interests. The tool first launched for the reels feedin December 2025, but is now available across the feed, explore, and reels. Once you access the tool in your settings, you see the topics Instagram thinks you care about the most. You can then tell the app what you’re interested in and what you want to see more or less of, and your recommendations will adapt accordingly. Instagram headAdam Mosseri has saidthat social media ranking models have historically been built with technology that wasn’t transparent to users, but now large language models (LLMs) can make recommendation systems more understandable by showing why content is displayed and letting users explicitly communicate their preferences. TikTok features a“Manage Topics” toolthat gives you more control over what you see in your “For You” feed. Launched in 2024, you can access the feature in your settings to customize your preferences for different topics, such as sports, travel, humor, current affairs, dance, and food. You can do so by moving the slider to adjust how much you do or don’t want to see that sort of content in your For You feed. If you are uncertain about which type of videos fall under a category, you can click the “information” button next to a topic. For instance, TikTok says the “Creative arts” topic includes “painting, drawing, graphic design, and art-related tutorials.” In 2025, TikTok expanded its Manage Topics tool by introducingAI-powered Smart Keyword Filters, which automatically limit content containing related keywords, such as synonyms. For example, if you use the Manage Topics tool to filter out “remodeling,” TikTok will also filter out “renovation” and “renovations.”
View
Samsung Unveils AI-Powered Pet Health Monitoring Tool for Galaxy Smartphones at VivaTech 2026
Samsung announced a new artificial intelligence (AI)-powered pet care solution at VivaTech 2026 in Paris on Wednesday. The feature is being developed in collaboration with startup Lifet. Part of Samsung's broader connected healthcare vision, it would allow Galaxy smartphone users to monitor health issues in their pets using just a photograph. It is claimed to be capable of analysing photos for signs of potential conditions like cataracts, dental diseases, and patellar luxation.
View
Pinterest launches an experimental AI shopping app called ‘Ask Pinterest’
Pinterest on Wednesday announced a new experimental app called “Ask Pinterest” that will allow the company to explore a more conversational approach to shopping and product discovery that could eventually find its way to the main Pinterest app. It also introduced other AI initiatives, including Pinterest Model Context Protocol (MCP), designed for advertisers running campaigns on Pinterest’s platform, and other AI ad tools. The news comes just ahead of the adtech industry’s annual gathering at Cannes Lions, which is this year mainly focused on how AI can serve the needs of advertisers and marketers. The “Ask Pinterest” online application gives the company another way to utilize its “Taste Graph” — its internal data mapping people to their interests and aesthetics. It will initially be available in limited access, the company said. The AI-powered experience is designed to expand the visual discovery experience Pinterest is known for beyond the main app to a conversational, chatbot-like interface where consumers can ask questions using natural language to get more personalized recommendations and inspiration. It also arrives as AI chatbots are increasingly competing with traditional search engines for consumers’ attention. Google has already put AI to work tohelp online shoppersfind what they need, track prices,and check out.ChatGPT alsoexperimented with agentic shopping, as haveMeta,Shopify, and others. Rather than turning itself into a source of product recommendations that other AI services could leverage through licensing deals, Pinterest has largely focused on using its own datato train AI modelsandpower its AI products. Plus, by making Ask Pinterest a standalone app, the company has a way to experiment with the technology without disrupting the main Pinterest experience. The company explains that Ask Pinterest could work for more complex or multi-step queries that wouldn’t fit a traditional Pinterest search. For instance, you could use the app to ask for help planning a dinner party or furnishing a room over time. The idea, says Pinterest, is to test and explore how AI could better support people’s shopping experiences while retaining the user’s context across sessions. Ask Pinterestcan also leverage users’ own saved Pins and Boards to personalize its answers when users sign in. In time, these results will help Pinterest when building more AI-powered experiences for the company’s flagship app, the company believes. Pinterest’s new app was announced alongside the updates aimed at marketers, including the introduction of an AI assistant, still in beta, in its Ads Manager in the U.S. A new AI model, Performance+ creative, was also introduced globally to help advertisers pick between different ad creatives to find the one that’s likely to perform best each time the ad is shown. And the MCP infrastructure layer that Pinterest announced will allow advertisers to manage and monitor their campaigns using other third-party agentic tools in a standardized way. In anannouncementsharing the news, Pinterest’s Chief Business Officer, Lee Brown, gestured towards the changing nature of web search, remarking that, “the future of discovery won’t be driven by keywords alone. It will be shaped by context, taste, and trusted recommendations” — an area where Pinterest feels it has a “unique advantage,” Brown said. Ask Pinterest is currently available to a limited number of people via the web, both mobile and desktop, the company says.
View
DeepL acquires Mixhalo for live-event audio streaming and translation
At conferences, speakers are often delivering their keynote or panel discussions in languages that many attendees might not know. That leads to users scrambling for their phones and opening translation apps to capture audio from a distance, which is not always effective. Mixhalo,a real-time audio startup that solves for situations like these, is joining DeepL to boost the German startup’s translation suite to help improve these kinds of translation experiences. Mixhalo was founded in 2016 by Incubus guitarist and songwriter Mike Einziger, violinist Ann Marie Simpson-Einziger, and Vik Singh, who is now the CEO of the startup. The company’s initial pitch was to improve the listening experience for concert attendees through its platform, but over the years evolved into a company that powers real-time audio for sports and live events. The startup raised over $39 million in capital from investors including Fortress Investment, Founders Fund, Defy Partners, and Cowboy Ventures. Mixhalo’s CEO Singh said in an email that tons of voice models coming to market were beneficial to Mixhalo, as it could integrate a variety of them and compare performance. He said that the rise of voice AI didn’t directly contribute to acquisition talks, but as model companies grow big, they would “start encroaching” on the space Mixhalo operated in, making it difficult to win on pricing. Mixhalo said it already relied on DeepL as its primary translation provider, and it made sense to work closer with the company. Singh tells us: “The DeepL conversation was very organic. Mixhalo has been a long-time DeepL customer, and I attended a customer dinner and ended up seated next to Sebastian, DeepL’s CTO. We just got to talking, and the more we talked, the more obvious the overlap became across the event space, the API, and the application layer, whether that is voice for meetings, document translation, or live event.” DeepL has been a text translation player for a long time, but in the last few years, it has started making noise about its voice products. In 2024, the company launchedvoice-to-text translation capabilities in over 33 languages. This April, it launcheda voice-to-voice translation suiteto support use cases like multilingual meetings. Mixhalo’s acquisition can push DeepL into the live event space with the same suite. “For us, Mixhalo will work as a solution and also a marketing use case. The platform will allow us to show how DeepL’s tech works in real-time and in environments like conferences where people are present on the ground,” DeepL CEO Jarek Kutylowski told TechCrunch in a call. Kutylowski said that with the acquisition of Mixhalo, which is based in San Francisco, DeepL is opening an office in the Bay Area to expand its U.S. operations. Mixhalo competes with the likes ofWordly AIandSeven Seven Six-backed Palabra.
View
