🚀 Zaprep: Your Socials on Steroids. 免费开始 — 每月自动发送 1,000 条私信，将互动转化为潜在客户。 ，每月 1,000 条自动私信。
最新 AI 资讯
OpenAI’s New Chip Beats NVIDIA; Gen 2 Nears Tape-Out in Months
“Jalapeño is also evidence of a broader full-stack advantage,” OpenAI said in the blog post.
View
India's Pest Surveillance AI Has Reached National Scale, But Its Impact Is Unproven
Powered by tech from Plantix and Wadhwani AI, the National Pest Surveillance System (NPSS) is an AI platform that detects crop pests and supports farmers.
View
Robotics startup Generalist reaches $3B valuation, sources say
Generalist, a robotics startup, is now valued at $3 billion after raising additional capital led by 8VC, according to two people with knowledge of the funding. The fresh capital totals nearly $200 million according to a regulatory filing. That additional capital is an extension of a $400 million Series B led by Radical Ventures that the company announced in June at a $2 billion valuation, the people said. The new capital brings the round’s total funding to $600 million. Generalist and 8VC didn’t respond to a request for comment. Generalist was founded in 2024 by former Google DeepMind researchers Pete Florence and Andy Zeng, along with former Boston Dynamics engineer Andrew Barry. It received early backing from 8VC and Radical Ventures as well as Nvidia, Union Square Ventures, Bezos Expeditions, and AI researcher Fei-Fei Li. Until recently, the startup operated quietly and with little publicity. Generalist is developing an AI foundation model that can work with various robots. It claims its newly released Gen 1.5 model enables robots to master new tasks from video demonstrations as short as 3 to 12 seconds long. The startup is working with a handful of customers, using their feedback to tailor the model for specific use cases, according to one source. Generalist isn’t alone in its pursuit of building a brain for a broad range of robots. Other competitors include Physical Intelligence, which is reportedly valued at $11 billion, and SoftBank-backed Skild AI, valued at $14 billion, as well as Genesis AI, which was in talks as of last month to raise capital at a $3 billion valuation. The funding surge reflects a bet from some investors that robotics may soon reach its own “ChatGPT moment,” meaning that robots will be able to perform general tasks without being explicitly trained for each one. However, because robots cannot be trained on the entirety of the internet’s data the way LLMs can, some VCs warn that a truly general robotics model may still be years away.
View
India’s Ringg gets backing from Peak XV as it pushes voice AI past the phone call
More than76% of consumersin India prefer talking to businesses over a phone call, according to a recent study from Truecaller. Since voice is still consumers’ preferred way to communicate, that leaves a big opportunity to automate support and outreach calls using voice AI in the country. Voice AI startup Ringg, which already processes 20 million call attempts a month, is betting that volume keeps climbing over the coming months, and it just raised more money on that belief. The company said today it has landed $10 million from Peak XV Partners as an extension of its Series A. It had previouslyraised $5.5 millionin a Series A round earlier this year, bringing the round’s total to $15.5 million. Ringg started life as a text-to-speech startup called DesiVocal, but training its own speech models proved expensive, so the founders moved up the stack — building voice AI agents for enterprises instead. Indian fintech Cred became its first customer, and Ringg has since signed Indian startups like Flipkart, Practo, Groww, and PolicyBazaar. “At the start, we were doing high-volume, low-complexity use cases like outbound calling, lead qualification, loan collection, and more. We quickly realized these are not sticky use cases, and so it’s always going to be a price game,” the startup’s co-founder Siddharth Tripathi told TechCrunch. Ringg still serves some of those simpler use cases, but it has set its sights on more complex workflows: appointment booking for healthcare clinics, abandoned-cart recovery for e-commerce sites, and onboarding/KYC (“know your customer”) checks for fintech apps. Tripathi said Ringg’s voice agent now runs across 1,200 clinics for the healthcare app Practo, helping patients book visits or follow up on next steps post-visit. Voice calls still make up over 70% of Ringg’s business, but the startup has started branching into other channels, including chat and WhatsApp. For some clients like Shell, it’s also automating browser-based support requests. “We are trying to position ourselves as a platform for agents that bring outcomes or get things done rather than voice agents for enterprises,” Tripathi said. Most of Ringg’s customers are based in India, with a handful in the Middle East and the U.S. But the startup isn’t trying to sell directly to U.S. companies; instead, it wants to partner with so-called Global Capability Centers in India — the offshore hubs multinationals increasingly lean on for back-office and support work — to sell automation capacity alongside human support. Tripathi said the company builds its own speech recognition and generation models, and would eventually like to own the full voice stack, including infrastructure and deployment. For now, though, that remains too costly, so the product works as an orchestration layer, routing tasks to different models depending on the use case. Rishen Kapoor, a principal at Peak XV, said that because Ringg started as a research lab building its own models, that technical depth shows up in the complex use cases it’s now tackling. “Because of the technical capabilities, they can actually do these hard-won enterprise workflows end to end. They can complete these higher-value tasks like merchant onboarding, like L1 and L2 support, with quality and with consistency,” Kapoor told TechCrunch. Voice AI in India is a crowded field. Model makers includingDeepgram,ElevenLabs,Cartesia, and local players likeSarvamandSmallest.ai, are all jockeying for pole position. Orchestration-focused startups like Bolna and Blue Machines are chasing the same layer Ringg occupies, while sector-focused players like Gnani and Arrowhead concentrate heavily on finance. That layered stack — model makers, orchestrators, and application-layer players all trying to lock in enterprise workflows — is itself the story. The money and the defensibility increasingly sit with whoever owns the customer relationship and the outcome. Ringg currently has 40 employees, with more than 15 hired in the last three months. The startup is hiring for forward-deployed engineer roles that combine technical chops with product management skills, along with researchers focused on bringing down the cost of running its models.
View
Shopify CEO Tobi Lütke Considers Banning Claude Code
“Insisting on only reading CLAUDE.md sometimes leads to split-brain problems when different team members use different tools.”
View
OpenAI loses a top data center exec, as stream of high-profile departures continues
OpenAI has lost yet another executive, and the timing of this one stands out in particular given this individual oversaw the execution of the company’s data center strategy. Chris Malone, OpenAI’s former head of data centers, left the company last week, the Wall Street Journalhas reported. Malone, whospent nearly five yearsat Meta and more than a decade at Google before that, joined OpenAI in March of last year, making his tenure relatively short. Malone joined the company not long after the launch ofthe Stargate Project, a $500 million data center initiative championed by the Trump administration that has sought to develop data centers in the U.S. OpenAI, along with Oracle, Nvidia, SoftBank and Microsoft, is considered a key partner in the effort. Why is Malone leaving? It’s isn’t entirely clear. With the AI infrastructure buildout frenzy in full swing across the industry, data center strategy has become one of the most closely watched roles at any AI lab, which makes turnover in that seat especially surprising. In a statement to TechCrunch about Malone’s departure, OpenAI said it had “recently reorganized” its “infrastructure organization to support the scale and pace of our work.” It added: “We have a strong, deeply experienced data center team in place, with clear leadership and the technical expertise to execute our plans.” As part of that reorganization, per the WSJ, Malone stopped reporting directly to OpenAI President Greg Brockman and began reporting to OpenAI Vice President Sachin Katti, who took over leadership of the group. Several other executives are said to be currently overseeing OpenAI’s data center strategy, including Uday Ruddarraju, who leads the company’s data center team; Brent Mayo, who leads the data center build and delivery program; and Spas Lazarov, a veteran of the data center and energy industries, who leads all data center engineering. Whatever the reason for Malone’s departure, he isn’t alone. His exit adds to a string of more than a dozen executive departures this year. Business Insiderrecently talliedthe total 2026 departure count at 13, with several leaving in just the last month. And these aren’t junior employees walking out the door — they’re some of the company’s most senior seats. Two weeks ago, the company replaced its chief revenue officer, Denise Dresser, after Dresser had only been with the company for some eight months. Two days prior to Dresser’s announced departure, the companyalso lost Brad Lightcap, one of its longest-serving executives, who spent years as the company’s chief operating officer. Lightcap said he would be “starting something new” but hasn’t given details on what that new project entails. Approximately a month before those departures, the companyalso lost Fidji Simo, the company’s de facto second-in-command (Simo served as product and business chief and reported directly to CEO Sam Altman). She stepped down from her position to recover from a “chronic illness.” Simo remains with the company in an advisory role. OpenAI’s safety and ethics teams have seen notable departures, too. In July, the companylost its “head of ethics,”Chloé Bakalar, and, last week,it was reportedthat the company had disbanded its preparedness team, a unit dedicated to assessing whether the company’s AI models could result in catastrophic risks. Other team leaders,like Bill Peebles, the former head of OpenAI’s now defunct AI image generator Sora, have left because their projects were shut down. In April, the companyalso lostits chief marketing officer, Kate Rouch, who — like Simo — reportedly left for health reasons. The company’s remaining talent has tried hard to minimize the significance of the ongoing talent flight, with co-founder Greg Brockmanrecently lamentingthat the intense “spotlight” on his company means that “every departure gets scrutinized in a way that it doesn’t otherwise.” But the churn has naturally raised questions, especially as it preps for an IPO. OpenAI’s public listing,originally expectedthis year, is now reportedly pushed to 2027, and the company is undergoing a kind of reputational vetting that comes with any looming listing. Concerns have arisen that the companymay be overvaluedand thatits profitabilitydoesn’t match the gargantuan investments being made in the lab. Suffice it to say, the flood of outgoing executives certainly hasn’t quieted those doubts.
View
Claude Cowork finally remembers what you told the app in chat
A change to Claude’s memory system will eliminate one of the most annoying things about using agents — the constant need to rebrief the AI on things it already knows. On Tuesday, Anthropic announced it’s merging the memory system used by chat and Claude Cowork, which means that Claude will always remember what it learned in one area, even when you’re engaging with it in another. Anthropic is also exposing what Claude has retained in its memory, allowing users to read, edit, or delete information on any topic. The update will improve the experience for Claude’s users as it will allow them to more seamlessly move between the research phase and taking action. Before, you may have developed ideas for projects over time while chatting with Claude, but when it came time to actually enact the changes, you’d have to start over by telling Cowork all the details. This was incredibly frustrating. The update makes Claude now feel like one continuous assistant, rather than two separate products under one roof. Anthropic offers an example of someone needing to draft an update for their manager or an agenda for a conference, where Cowork would already understand the references because of the past conversations about the matter at hand. For instance, it would know the headcount, city, and speakers at the conference, so you wouldn’t have to re-input that information. In addition, Claude will now add topics to its memory as you chat, instead of summarizing a conversation when it ends. That makes it easier to move back and forth between chat and Cowork, as the new memories are more quickly updated and shared between the two experiences, even if the chat is ongoing. Anthropic says that, by default, Claude won’t store personal or sensitive information like your health data, race, ethnicity, religious beliefs, politics, gender identity, and more. However, users can opt to enable these things by toggling on “include sensitive topics in memory” if they prefer. The app will also notify users whenever a sensitive topic is saved in memory. However, the company says that Claude won’t ever save certain things, like government-issued ID and Social Security numbers, criminal history, immigration status, or other things that violate its acceptable use policy. The memory feature is enabled by default on Free, Pro, and Max plans across web, desktop, and mobile. On iOS and Android, users will need to update their app to the latest version.
View
Stability AI, maker of image generator Stable Diffusion, raises $76 million in fresh funding
Stability AI, the startup behind AI image generation model Stable Diffusion, has raised $76 million in Series B funding. The new haul brings the company’s fundraising total to $232 million. The companyannouncedthe round Tuesday, saying the new capital comes partly from a number of prominent entertainment industry organizations, including Universal Music Group, Sony Music Group, and Warner Music Group, and gaming giant Electronic Arts (EA). Two investment firms, AMD Ventures and Pacific Alliance Ventures, also took part. It’s an unusual roster and less a typical venture round than a lineup of the companies Stability now depends on for content licensing and distribution deals. Stability AI CEO Prem Akkaraju, who joined the company in 2024, called the funding “an affirmation of our vision where generative AI empowers every producer, musician, and storyteller.” Akkaraju joined the company in 2024. The company says it plans to use the money from its latest funding round to continue building out its “creative production” product suite while also expanding its professional services arm. Currently, Stability offers a variety of AI models that are designed for the purposes of AI music, video, and image generation. Stability, founded in 2019, has spent the past year signing deals to weave generative AI into entertainment companies’ creative workflows. It struck partnerships withUniversal MusicandEAlast October and withWarner Musiclast November, each giving those companies a hand in co-developing Stability’s AI tools rather than just licensing the output.. Stability has also had a mostly winning stretch in court. The company largely prevailed ina copyright lawsuitbrought by Getty Images in the United Kingdom, which had accused the company of infringing upon IP rights by using the company’s images in the training of its image generation model. A judge ruled largely in Stability’s favor in that case, but a similar lawsuit from Getty in the U.S. is stillworking its way through the courts. Stability was also sued in 2023 by the company’s co-founder, Cyrus Hodes, who claimedthat he was trickedby the other co-founder, Emad Mostaque, into selling his share in the company.
View
Accel-backed Keenable is indexing the web for AI agents
Search engines were built and optimized for people, who can’t spare the time or attention required to scan entire webpages. But as people increasingly use AI chatbots to search the web and do tasks, there’s a line of thinking that the Internet’s current infrastructure needs to be updated to cater to AI instead, as these bots can read and process much larger portions of information. Andrey Styskin, who previously led Russian search giant Yandex’s search, AI and cloud division, and German AI scientist Matthias Petri are working to solve that problem with their new startup,Keenable. The company recently came out of stealth $26 million in seed funding. Accel led the funding round, which also saw participation from Conviction Partners and some business angels. From Styskin’s perspective, AI chatbots tend to do much better if they canground their responseswith source documents. “This actually creates a new flywheel that is different from what Google learned from human behavior,” he told TechCrunch. Keenable says it has been building a web search index of more than 100 billion documents, and its API is already used in production at several AI labs and inference providers during both training and runtime. The startup wouldn’t disclose who its customers are, but it recently struck apartnership with voice AI company Gradiumto support live information retrieval. Drawing from his experience of 20 years building search at Yandex and Amazon, Styskin explained how Keenable’s product is different from enterprise search solutions that can break down and prove very costly at web scale. “If you do not fine-tune your index structures for a specific task, the cost of serving and scanning the whole internet is enormous because of the volume. That’s why you need to innovate on how you can narrow the search space based on your query very fast. This is what we are bringing to the table,” he said. According to Accel partner Zhenya Loginov, who led the investment, AI players have very few options when it comes to web-scale search infrastructure, especially with Google and Microsoft taking steps toshut downtheirexisting search APIsto avoid cannibalization. Instead, the tech giants are opting for a more bundled approach, and being selective about their partners. To Styskin, these decisions confirmed the opportunity he saw when he was at Amazon, working with Petri on web search infrastructure for AI applications such as Alexa. He’d seenCloudflare datathat AI crawlers were responsible for a growing share of search volume, and started realizing that there was an opportunity to develop web search infrastructure that is built with AI in mind. Armed with that experience, Styskin tapped into his network to hire a handful of former colleagues for his new startup, which is also building proprietary retrieval capabilities. This includes an upcoming product, Web Query Language, which would help AI systems answer questions by combining information from various web sources, even when none contain the full answer. Costs will be an important part of the equation. Styskin says while it is “extremely hard” to convince people to move away from Google for search, theinnovators’ dilemmameans that the U.S. giant is potentially “beatable” on agentic queries, and a smaller company like Keenable can innovate and offer a more cost-efficient solution to AI companies. Still, the costs of building a giant search index are real. “Don’t ask — it is painfully expensive,” he said. But, he says the startup is doing its best to keep costs in check and pace itself. With a team of 15 engineering staff across the U.S. and Europe, the company plans to use the fresh cash to double its headcount by the end of the year to build its go-to-market motion. There are many more steps before the startup can achieve its dream of becoming “the next Google for AI agents.” Other players have entered the space, such asBraveandExa; and Google itself is overhauling its search experience for the AI era. But this broader motion indicates that Keenable’s conviction is also shared at the Googleplex: whether it’s for humans or for agents,the era of the “ten blue links” may be coming to a close.
View
OpenAI’s Jalapeño chip is built for fast inference at scale, benchmarks show
At the Hot Chips conference on Tuesday, OpenAI shareda more detailed look at Jalapeño, including the first batch of benchmark results for the new system. Tested on SemiAnalysis’ InferenceX benchmark, Jalapeño registered both more tokens per user and more throughput per kilowatt than the currently available state-of-the-art inference processors. “The bottom line is that the results show a very, very significant performance advance over state of the art,” said Richard Ho, OpenAI’s head of hardware, in a press call. “Jalapeño can serve more AI work per unit of power, while also returning responses more quickly. It’s very efficient to serve a lot of customers, but it can also be very low latency.” Notably, that comparison is against an Nvidia Blackwell system — but by the time Jalapeño reaches full deployment, the competition may have advanced significantly. Ho estimated that Jalapeño would deploy at the end of 2026 “in very small volumes,” with more significant deployment coming in 2027. First announced last October, Jalapeño was developed by OpenAI in close collaboration with Broadcom, with OpenAI’s own models assisting in the development process. The company plans to make Jalapeño a multigenerational platform, allowing AI products, models, chips, and memory all developed in concert. Because of that full-stack approach, OpenAI was able to address specific phases in the inference process that often cause friction during inference processing. In particular, Jalapeño is designed to minimize delays during the prefill and communication phases of processing, which OpenAI says often act as bottlenecks. “We designed Jalapeño to minimize data movement and communication delays,” the company said in a blog post presenting the results. “This means that model state, including the KV cache used while generating a response, can be explicitly placed and kept local while the system activates the right combination of compute, memory, and networking for each inference phase.”
View
Gamma acquires Accel-backed design startup Lica
Presentation startupGammahas acquired Accel-backed design startupLicain a bid to build out its own design research lab, TechCrunch has exclusively learned. Lica’s co-founders will lead the effort. Lica started off as an app that could turn screenshots and recordings into project presentations and videos. Founded by Priyaa Kalyanaraman and Purvanshi Mehta in 2023, the company a year later raised $4 million from investors including Accel, South Park Commons, and Village Global. Over the past couple of years, it has focused on working with e-commerce sites to create marketing videos that adhere to brand guidelines. Kalyanaraman said Lica wanted to build AI models around communication methods, including video and design, and the founders were familiar with Gamma’s CEO, Lee Grant, as both companies are backed by Accel and South Park Commons. Those conversations eventually led to the acquisition. “We have a very similar North Star vision of wanting to make visual communication easy. We focused a lot on frontier research and making sure that became accessible to users. But that’s just one part of the solution. Gamma had focused a lot on building distribution, taking the application to 100 million-plus users. Given how fast things are moving, we decided to join forces,” Kalyanaraman told TechCrunch. Gamma will continue to focus on helping people build presentations, but like other design companies, it’s also started offering image generation. The company says it now wants to explore different formats of communication with the new research division. “Presentations for us are today a core use case, but how do we expand what presentations look and feel like, and how do people engage with them in the future? What visuals are possible? How interactive are they? These [questions] all require us to think deeply about the core responsibility of communication. This is why we’re excited to invest in this area, because we don’t think anybody else is thinking at that level,” Lee said. Lica’s Mehta said the new research division would work on customizing the presentation and communication styles of AI models for different audiences and tastes. “The point of the presentation is to communicate a large goal. We are thinking about building a platform where these two goals can be personalized to such an extent that you just get a really good output, and then you’re able to edit for different audiences,” she said. Gamma didn’t reveal what products it plans to build as a result of this acquisition, but it did say that it wants to create fluid, multimodal presentations and other communication modalities. AI presentation startups have received tons of investor attention and money over the last few years. Prezent raised$50 million in two rounds; Gamma raised$68 million at a $2.1 billion valuation; andPresentations.ai was also backed by Accel. We’re even starting to see some consolidation in this category: Earlier this month,OpenAI acquired NextSlide.
View
Why India’s IT Giants Need Mexico When AI Shrinks the Team
Indian IT services firms are using Mexico for customer-facing work as AI reshapes global engineering and delivery.
View
