Latest AI News
Lovable says it added $100M in revenue last month alone, with just 146 employees
Lovablecrossed $400 million in annual recurring revenue in February, the Stockholm company confirmed to TechCrunch. But it declined to say whether it is stillprojecting to reach $1 billion ARR by year’s end, saying its focus is on “helping builders scale their impact with our platform.” Alongside Cursor, Mercor, and others, Lovable is part of a wave of tools that make it easier to create websites and apps using natural language, a practice known as vibe coding. This initially resonated with individuals and startups, but the three-year-old company has been pushing hard to secure enterprise clients, which already include Klarna, HubSpot, and others. Lovable’s debut brand campaign, “Earworm,” which began running this week across social platforms, YouTube, and connected TV, still speaks to mainstream users. The film follows a woman who can’t rid herself of a song — performed by Swedish bandBoko Yout— until she finally opens Lovable and builds it into a working app. The creative team behind the campaign built the band app that’s featured in the film using Lovable itself as a functional, live product, in fact. “The purpose of this brand campaign is to inspire the next generation of builders — non-technical people with great ideas that deserve to come to life,” a spokesperson told TechCrunch. That overarching message is one of the factors that have helped Lovable attractsome 8 million usersandbecome a unicornin less than a year after its launch. But the prospect that it could also secure enterprise dollars likely played a key role in boosting its valuation to$6.6 billion. More than half of Fortune 500 companies are using Lovable to “supercharge creativity,” co-founder and CEO Anton Osikadeclared at Web Summit last November. The company has added a range of dedicated features — often security-related — to convince businesses to use it for more than prototyping and keep them from canceling over time. Disclosing ever-increasing ARR numbers is also a way for the company to show its success is not fading. It had previously reported$100 million ARRlast July,$200 millionlast November, and$300 millionin January, which suggests that its revenue growth has been accelerating in recent months despite the rise of AI coding tools from major AI labs like Anthropic and OpenAI. Neither Claude Code nor Codex is a vibe-coding platform, and the notion that they could create full apps as seamlessly may be overrated, but their parent companies may eventually decide to compete with Lovable, which is built atop their models. However, Osikahas shown little concern, and the company’s latest usage metrics offer some support for that confidence. Its most recent user spike was tied to a specific promotion — Lovable’s SheBuilds initiative for International Women’s Day on March 8, when the whole platform was free for one day. “We saw various records set,” the company told TechCrunch. “One we’re most proud of is that over 500,000 projects were built or updated on Lovable that day (compared to a typical daily average of [approximately] 200,000).” Also notable is the fact that Lovable achieved $400 million ARR with only 146 full-time employees, as chief revenue officer Ryan Meadowstold Business Insider. The company now plans to increase its headcount — and there’s room for it. Its recently inaugurated space in Stockholm hasspace for 300 people, and the company is also hiring in Boston, London, New York, San Francisco, and remotely. Even accounting for these70 open positions, Lovable’s revenue-to-employee ratio will likely remain well above industry norms. Research firm Gartner predicts that a new wave of unicorns will emergeby 2030 with $2 million ARR per employee. At $2.77 million in ARR per employee, Lovable has already surpassed that number.
View
Netflix may have paid $600 million for Ben Affleck’s AI startup
Last week, Netflixannouncedthe acquisition of InterPositive, an AI company co-founded by Ben Affleck that helps filmmakers edit footage in post-production. This deal could be worth up to $600 million,according to Bloomberg, potentially ranking it among the streaming giant’s largest acquisitions ever. The most Netflix has ever paid for a single acquisition was approximately$700 millionfor the Roald Dahl Story Company. While Netflix has not publicly confirmed the details, sources tell Bloomberg that the actual cash payment may be lower, with the owners of InterPositive eligible for additional payouts tied to specific performance targets. InterPositive makes tools that help filmmakers work more efficiently in post production — addressing continuity issues or enhancing scenes, for example — but it doesn’t generate new content or use footage without permission. This acquisition fits Netflix’s broader push to integrate AI into content production. The company has alreadyused generative AIin its original shows and movies, including to create a building-collapse scene in the Argentine series “The Eternaut.” Rivals are moving in the same direction.Amazonis building in-house AI teams for film and TV projects, whileDisneyhas struck a deal with OpenAI. Not everyone is on board. Workers across the film industry have raised concerns about potential job losses and whether AI companies are compensating creators fairly for training data.
View
Zendesk acquires agentic customer service startup Forethought
Zendesk is acquiring Forethought, a company that builds software to automate customer service interactions, thecompanies announced on Wednesday. The deal is expected to close by the end of March. While AI agents, particularly for customer service, may be all the rage now, Forethought was years ahead of its time. In fact, it wasthe 2018 winnerofTechCrunch Battlefield, our iconic startup pitch competition. For context, ChatGPT didn’t launch until late 2022. Following its Battlefield win, Forethought landed marquee customers like Upwork, Grammarly, Airtable, and Datadog, it said, and by 2025 was supporting more than a billion monthly customer interactions. Terms of the deal were not disclosed. Forethought raised total funding of $115 million from backers like Blue Cloud Ventures, NEA, Industry Ventures, Neo, Village Global, and Sound Ventures, as well as angels like May Habib (Writer), Scott Wu (Cognition), Karan Goel (Cartesia), and Gwyneth Paltrow, the company said when itraised a $25 million round last year. Forethought AI co-founder and chairman Deon Nicholas called the acquisition a milestone in aLinkedIn post. “More than seven years ago, we set out with a simple but ambitious idea: AI could transform the customer experience. When we first launched Forethought at TechCrunch Disrupt, that vision felt bold—even a little crazy. Today, AI agents aren’t just transforming customer experience. They’re transforming every industry imaginable,” he wrote. Shortly before the acquisition was announced, Nicholas appeared onTechCrunch’s Build Mode podcast, where he discussed how he prepared to win Battlefield, how he landed his first customers, and what comes next for agentic tech, like browser control. By the way,applications for the 2026 TechCrunch Startup Battlefield are currently open. Zendesk, best known for its self-help customer service products, says it will continue to support Forethought’s existing customers and integrate the startup’s technology into its own AI products — including more specialized agents, self-improving AI, voice automation, and more autonomous capabilities. The company says the acquisition accelerates its product roadmap by more than a year. Zendesk has been privately held since November 2022, when it was acquired by a consortium led by private equity firms Hellman & Friedman and Permira in a$10.2 billion deal. Not disclosing terms of this newest deal is pretty consistent with its pattern over time. Zendesk has made roughly a dozen acquisitions since its founding in 2007, and the few times it has disclosed a price, the figures were modest, including$29.8 millionfor live-chat company Zopim in 2014, and$45 millionfor analytics firm BIME in 2015.
View
Ford’s new AI assistant will help fleet owners know if seatbelts are being used
Ford rolled out an AI assistant this week that can monitor and analyze millions of data points to help its Ford Pro commercial customers boost their bottom line. The bet, and one that most other automakers are making, is that there’s money to be made in software. Ford Pro AIdebuted at Work Truck Week in Indianapolis and is now available to all of its U.S.-based Pro telematics subscribers. The AI assistant is included in the subscription. Ford doesn’t disclose how many U.S. subscribers it has; it has more than 840,000 global subscribers. Ford Pro, which generated $66.3 billion in revenue in 2025, is a sensible target for the company as it seeks ways to give its paying customers more value. But it’s not its only one. Fordannounced earlier this yearat CES 2026 that it’s developing an AI assistant for owners of its passenger cars and trucks that will debut in the company’s smartphone app, before expanding to its vehicles in 2027. Ford emphasized to TechCrunch that this is not a mere chatbot. Instead, the company said its proprietary systems give subscribers detailed information about fuel consumption, seatbelt use, and vehicle health, not just a bunch of diagnostic error codes when something is wrong. It can also provide managers with information on idle times, speeding, and acceleration events across the fleet. Like its consumer AI assistant, Ford Pro AI is built off of Google Cloud and uses a number of AI agents. The secret sauce, per Ford, is its use of internal data from each customer’s fleet to reduce the potential of AI hallucinations and errors. Ford Pro, a business division that sales to Super Duty large trucks as well as commercial, government and rental customers, has become a moneymaker for the automaker. The Ford Pro business division reported a net income of $6.8 billion in 2025,according to its earnings report. The company said Ford Pro paid software subscriptions grew by 30% in 2025. Even as Ford rolls out AI tools for its customers, executive leadership has warned of impending job cuts because of the technology. Last year, CEO Jim Farleypredicted AIwould halve the number of white-collar jobs in the United States. In January, Farleysaidthat the U.S. needed essential workers to build and support the infrastructure needed to reach its AI moonshot goals.
View
Replit Raises $400 Mn at $9 Bn Valuation, Unveils Agent 4 for Vibe Coding
Replit is on track to reach $1 billion in ARR by the end of the year.
View
Perplexity AI Launches Personal AI Computer That Works 24/7 on a Mac Mini
It acts as a digital proxy that can access files, manage tasks and coordinate tools across devices.
View
Meta’s Moltbook deal points to a future built around AI agents
When news broke Tuesday morning thatMeta bought Moltbook, the social network for AI agents, it may have left some people scratching their heads. What on earth would Meta — an ad-supported company — want with a social network where the users are bots? Bots, after all, are not the target audience of brand marketers and advertisers. Meta isn’t saying much. Its only official comment was a brief statement that the Moltbook team was joiningMeta Superintelligence Labs, which would open up “new ways for AI agents to work with people and businesses.” Reading between the lines, this was an acqui-hire. A network built for bots isn’t exactly a natural home for brand advertising — even if Moltbook was never entirelynon-human. What Meta really wanted was the talent behind it — people who are having fun brainstorming and experimenting with AI agent ecosystems. And that, counterintuitively, could be a boon for its advertising business. As Meta CEO Mark Zuckerbergsaid last year, he believes in a future where “every business will soon have a business AI, just like they have an email address, social media account, and website.” On an agentic web, one where AI systems act independently on users’ behalf, AI agents could interact with each other, doing things like buying ads, making bookings, and responding to customers. AI is alsobeing usedto generate ad creative and tailor its output based on who’s viewing it. AI systems could also manage product pricing or generate personalized offers. On the consumer side, agents could be used tofindthe best prices and deals, manage bookings, andshopfor products. Insomelimitedcases,agentscan already check out and pay on consumers’ behalf. (Agentic commerceis still in its early days, and these systems don’t always work as well as advertised. But the market has been moving fast, and improvements seem likely soon enough.) As Facebook once built the “friend graph” — a network defined by social connections between people, where every individual is a node — an agentic web could benefit from an “agent graph,” a system that maps out how various agents are connected and what actions they can take on each other’s behalf. For an agentic web where businesses’ agents and consumers’ agents can work together, though, the agents first need to be able to find each other, connect, and coordinate their activities. As Facebook once built the “friend graph” — a network defined by social connections between people, where every individual is a node — an agentic web could benefit from an “agent graph,” a system that maps out how various agents are connected and what actions they can take on each other’s behalf. This could span areas like travel, online shopping, media and research, productivity tools, and more. This, too, could be where advertising slots in. Today, humans view and click on ads when they see something of interest, but on an agentic web where agents are shopping on users’ behalf, ads might look quite different. Instead of influencing a human to buy a product, a business’s agent may need to negotiate directly with a consumer’s agent to make the sale. Maybe the consumer wants to buy that shirt or that lipstick, but only in a certain color and at a certain price. Maybe the systems become so complex that these considerations go beyond product and price — perhaps the consumer prefers to support small businesses, or shops only with eco-friendly companies. Maybe the consumer only buys items when they’re on sale or purchases generic versions if the ingredients are the same. And so on. In that case, it’s not just a matter of connecting the AI agents but also ranking products by whichever one best fits that individual customer’s needs. If Meta could capitalize on that market — AI at the orchestration layer, meaning the system decides which agents talk to each other and in what order — it could potentially expand its ads business into entirely new territory. This all depends on whether consumers actually embrace the agentic web, or ever trust AI enough to let it act on their behalf. But the very existence ofOpenClaw, the personal AI assistant that populated Moltbook with content, suggests that at least some people are already leaning into autonomous AI agents. Of course, there’s another possible reason Meta bought Moltbook. The companylost the acqui-hireof OpenClaw’s creator, Peter Steinberger, to rival OpenAI, so it went after Moltbook, the platform Steinberger’s tool helped build, instead. Petty? Maybe. But it kept Meta’sSuperintelligence Labsin the news.
View
Rivian spin-out Mind Robotics raises $500M for industrial AI-powered robots
Mind Robotics, an industrial robotics lab spun out of the electric vehicle maker Rivian, hasraised$500 million in a Series A funding round co-led by venture firms Accel and Andreessen Horowitz. The financing, announced Wednesday, follows a $115 million seed round that was led by Eclipse in late 2025, bringing Mind Robotics’ total fundraising to $615 million in the few months since its founding. This round brings the startup’s valuation to around $2 billion, according to The Wall Street Journal, whichfirst reported the news. Mind Robotics was created by Rivian CEO and founder RJ Scaringe. It was spun out of Rivian in November 2025, with Scaringe serving as chairman. The general idea is that Scaringe wants to use data from Rivian’s electric vehicle factory to train industrial robots to be more dexterous and adaptable, as well as a venue to prove out those robots’ usefulness. The company “was founded to address a structural gap with current industrial automation solutions,” according to a press release announcing the Series A round. “Existing industrial robotics can perform repeatable, dimensionally stable tasks, but a large share of factory value-add work requires human-like dexterity, adaptation, and physical reasoning that classical robotics cannot address. Mind Robotics is building the AI foundation — models, hardware, and deployment infrastructure — to close that gap.” Scaringe told The Wall Street Journal that Mind Robotics will have a large number of robots deployed by the end of this year. In the months since Mind Robotics was announced, he has spoken a few times about how the startup intends to focus on more traditional factory robot designs, instead of the much-hyped humanoid robots that have garnered so much attention over the last year, likethose built by Tesla. “Doing cartwheels does not create value in manufacturing,” Scaringe told The Wall Street Journal. Beyond the training data and a place to deploy the robots, there are other ways Rivian and Mind Robotics might collaborate moving forward. In December, Rivian announced it had been developing its own custom silicon meant to help power the autonomous vehicle software that will go on its cars. In an interview with TechCrunch at the event, Scaringe said “it doesn’t take a lot of imagination” to think that Rivian might sell those custom chips to Mind Robotics. “It’s a robotics processor, so it could work really well for that,” he said. Mind Robotics is the second company Rivian spun out in 2025. Thefirst was Also, an electric mobility company that is starting with ahigh-end modular e-bike, as well as small electric cargo vehicles for Amazon. That startup was also backed by Eclipse, and has sinceraisedanother $200 million from Greenoaks Capital, with its valuation currently sitting around $1 billion.
View
WordPress debuts a private workspace that runs in your browser via a new service, my.WordPress.net
WordPress’s publishing software can now run entirely in the web browser, the organization behind the open source publishing softwareannouncedon Wednesday. Through a new service calledmy.WordPress.net, the WordPress software lets users set up a site and begin publishing without signing up, setting up a hosting plan, or registering a domain. Instead, the new solution leverages the same technology that powers WordPress demos and makes it available as a permanent, personal publishing platform. There is a big caveat to running WordPress this way: The sites set up on my.WordPress.net are private by default and not accessible from the public internet. “They aren’t optimized for traffic, discovery, or presentation, and they don’t need to be,” ablog postintroducing the new service explains. “Instead, WordPress becomes a personal environment where ideas can exist before they are ready to be shared, or where they may never be shared at all.” The sites created through this service are bound to your web browser, with their data saved in the browser’s storage. That means you can’t access the site from another device. But you can move your site to a dedicated WordPress host if you ever want to make it public. This positions WordPress as a personal workspace for activities like private writing, journaling, drafting, research, and learning, or building tools for personal use. For the latter, my.WordPress.net comes equipped with an App Catalog offering a variety of tools built with WordPress plug-ins, including a Personal CRM, Personal RSS Reader, a bookmarking tool, an AI Workspace, and more. The post notes that my.WordPress.net is powered byWordPress Playground, the open source project that lets you install WordPress on any device with one click and that integrates withOpenAIandCLI appsto create new tools. As a result, you can use an AI assistant to modify my.WordPress.net, to do things like tweak a plug-in or build a new one. You can also ask the assistant about data stored in WordPress, which it remembers, allowing WordPress to become a personal knowledge base accessed by AI. The service will take longer to launch the first time you use it, the post cautions, and recommends that backups should be saved regularly. Its storage starts at roughly 100MB, which makes it better for smaller, personal apps and use cases. If at any time you want to delete your current work, you can click a button to reset the site. Or you can set up new, temporary instances that reset themselves when the browser is refreshed. The launch of the service follows the formation ofa WordPress AI teamlast year, focused on launching new AI products for the developer community. The commercial hosting platform WordPress.com also launchedan AI website builderlast year that lets you design a site using an AI chatbot-style interface.
View
Replit snags $9B valuation 6 months after hitting $3B
Vibe-coding sensation Replit has hit another funding milestone. The companyannouncedWednesday that it raised a $400 million Series D at a $9 billion valuation, led by previous investor Georgian Partners. Other participating investors include G Squared, Prysm Capital, Coatue, Andreessen Horowitz, Craft Ventures, Y Combinator, Accenture Ventures, Okta Ventures, and Databricks Ventures. Founder and CEO Amjad Masad also saidin a post on Xthat backers include angel investors Shaquille O’Neal and Jared Leto. This followsthe $3 billion valuation the startup hit in September, when it raised a $250 million round. At that time, Replit said it was on track for $150 million in annualized revenue. Replit did not release updated, current ARR figures, but the company toldForbesit hopes to hit annual recurring revenue of $1 billion by the end of the year. While Replit’s recent rise has been meteoric,Masad previously told TechCrunchthat it took nine years of grinding — and a controversial decision to pivot from serving professional developers towards non-programs — to find its groove.
View
YouTube’s Likeness Detection Tool Expanded to Government Officials and Journalists
YouTube began rolling out its Likeness Detection tool to creators on the platform in October 2025. The tool was designed to monitor videos on the platform that mimic the likeness of face or voice of a user without consent. The feature was part of the Google-owned video streaming giant's initiative to protect users from the harmful usage of artificial intelligence (AI). On Tuesday, the company announced that the tool will now be expanded to civic leaders and journalists. They will have to first enrol themselves, and once done, they can begin requesting the removal of their deepfakes on the platform.
View
Canva’s New AI-Powered Magic Layers Feature Turns Images Into Editable Designs
Canva, on Wednesday, announced a new artificial intelligence (AI) tool for users. Dubbed Magic Layers, it can convert any flat or static image and AI-generated outputs into a layered, editable design. This means users can take any image and then manipulate its elements individually to create a different design. The Sydney-based visual communication platform stated that the tool was developed using an in-house foundation model. Currently, it is available in public beta in select regions, but it is expected to be expanded to more countries soon.
View
