AI 50 2025: AI Agents Move Beyond Chat
The 2025 edition of the AI 50 list shows how companies are using agents and reasoning models to start replacing work previously done by people at scale.
Artificial intelligence is entering a new phase in 2025. After years of AI tools mostly answering questions or generating content on command, this year's innovations are about AI actually getting work done. The 2025 Forbes AI 50 list illustrates this pivotal shift, as startups on it signal a move from AI that merely responds to prompts to one that solves problems and completes entire workflows. For all the buzz around big AI model makers like OpenAI, Anthropic or xAI, the biggest changes to the story in 2025 are in these application-layer tools that use AI to produce real business results.
Sequoia
Historically, AI assistants could chat or provide information, but humans still had to act on the output. In 2025, that's changing. For example, legal AI startup Harvey is showing that its software can do more than just answer legal questions–it can handle entire legal workflows, from document review to predictive analysis of cases. Its platform can draft documents, suggest revisions and even help automate negotiations, case management and client contact—suites of tasks that normally require a team of junior lawyers. This achievement earned Harvey a spot on the Forbes AI 50 list, and it exemplifies AI evolving from a helpful tool to a hands-on problem solver.
Many of the standout AI 50 companies are enterprise tools that do on-the-job work. Sierra and Cursor, the only product of startup Anysphere, are emblematic of this new generation of business AI. Sierra automates customer service while vastly improving the experience, opening the way for companies to offer help to their customers at all hours. Cursor, meanwhile, has taken the software developer community by storm. Its technology allows anyone to not only autocomplete lines of code (à la GitHub CoPilot), but also generate entire features and applications simply by asking for it in plain English.
Although we're not yet at a point where companies are deploying robots en masse, startups focused on robotics have made meaningful strides in the past year as they integrate models built using transformers (the 't' in ChatGPT) with hardware. Robotics featured prominently in Nvidia's keynote at its recent developer conference, with Jensen Huang claiming, 'physical AI for industrial and robotics is a $50 trillion opportunity.' Figure AI recently announced its BotQ high-volume manufacturing facility capable of producing 12,000 of its humanoid robots a year along with its new generalist Vision-Language-Action (VLA) model, Helix. Another AI 50 company, Skild AI, is taking a different approach, focusing on developing a universal robotics foundation model, Skild Brain, that can be integrated into a wide variety of robots, rather than building its own. Skild is also planning to sell services to the robotics industry that use Skild Brain.
So far, everyday consumers have mainly encountered advanced AI via chatbots like OpenAI's ChatGPT or Anthropic's Claude (and newcomers like Grok). Hundreds of millions have tried talking to these AI tools, but we've yet to see a truly mainstream app where AI completes everyday tasks for people. That is likely to change in 2026. With the technology maturing—and people realizing how AI can save them time and money at work—expect a wave of consumer-facing AI products that handle whole tasks on their users' behalf. For example, Anthropic recently launched Claude Code, making software writing accessible to consumers. These new capabilities raise the prospect that AIs able to manage your schedule, book travel or organize your files end-to-end are on the horizon.
This year is a turning point: AI graduated from an answer engine to an action engine in the workplace. The 2025 AI 50 companies proved that AI can be trusted with meaningful workloads and deliver real results, setting the stage for broader adoption. As we head into 2026, expect the enterprise gains to start spilling into everyday life. Challenges like accuracy and security remain, but the momentum is undeniable. In 2025, AI gets real work done.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
TechCrunch
18 minutes ago
- TechCrunch
Google reportedly plans to cut ties with Scale AI
In Brief Meta's big investment in Scale AI may be giving some of the startup's customers pause. Reuters reports that Google had planned to pay Scale $200 million this year but is now having conversations with its competitors and planning to cut ties with the startup. Microsoft is also reportedly looking to pull back, and OpenAI supposedly made a similar decision months ago, although its CFO said the company will continue working with Scale as one of many vendors. Scale's customers include self-driving car companies and the U.S. government, but Reuters says its biggest clients are generative AI companies seeking access to workers with specialized knowledge that they can use to annotate data for training models. A Scale spokesperson declined to comment on the company's relationship with Google, but he told TechCrunch that Scale's business remains strong, and that it will continue to operate as an independent company that safeguards its customers' data. Earlier reports suggest that Meta invested $14.3 billion in Scale for a 49% stake in the company, with Scale CEO Alexandr Wang joining Meta to lead the company's efforts to develop 'superintelligence.'
Yahoo
37 minutes ago
- Yahoo
Week in Review: WWDC 2025 recap
Welcome back to Week in Review! We have lots for you this week, including what came out of WWDC 2025; The Browser Company's AI browser; OpenAI's partnership with Mattel; and updates to your iPad. Have a great weekend! The Apple experience: We kicked the week off with WWDC 2025, Apple's Worldwide Developers Conference, where the company showed off a newly designed iOS 26, new features across its products, and much more. There was considerable pressure on Apple this year to build on its promises and to make amends to developers as it lags behind in AI and faces continued legal challenges over its App Store. Snack hack: U.S. grocery distribution giant United Natural Foods (UNFI) was hit by a cyberattack, the company confirmed Tuesday. Much of UNFI's external-facing systems were offline, including web systems used by suppliers and customers, as well as the company's VPN products. Whole Foods was one of the victims, and it told staff that the cyberattack was affecting UNFI's 'ability to select and ship products from their warehouses' and that this will 'impact our normal delivery schedules and product availability.' Public debut: Chime's much-anticipated public debut finally arrived, with the company raising $864 million in its IPO. Iconiq was one of Chime's many backers taking a victory lap at its graduation to become a public company. This is TechCrunch's Week in Review, where we recap the week's biggest news. Want this delivered as a newsletter to your inbox every Saturday? Sign up here. Not to be outdone: Google rolled out Android 16 to Pixel phones, adding group chat to RCS, AI-powered edit suggestions to Google Photos, and support for corporate badges in Google Wallet. Cabs are here: Elon Musk has spent years claiming that Teslas would be able to drive themselves. Apparently the time has come — maybe? Musk said this week that Tesla will start offering public rides in driverless vehicles in Austin, Texas, on June 22. An AI browser: The Browser Company said last year that it's going to stop supporting and developing its Arc browser, which, although popular, was never able to reach scale. The startup has since been busy developing an AI-first browser called Dia. And another one: OpenAI released o3-pro, which is a version of o3, a reasoning model that the startup launched earlier this year. As opposed to conventional AI models, reasoning models work through problems step by step, allowing them to perform more reliably in domains like physics, math, and coding. In other news, Sam Altman posted on X to say that his company's first open model in years will be delayed until later this summer. Desperately seeking: Now that people can ask a chatbot for answers — sometimes generated from news content taken without a publisher's knowledge — there's no need to click on Google's blue links. And that's hurting publishers. Cool? Mattel and OpenAI are teaming up to create an 'AI-powered product,' whatever that is. As part of the deal, Mattel employees will also get access to OpenAI tools like ChatGPT Enterprise to 'enhance product development and creative ideation.' 'A privacy disaster': Reporter Amanda Silberling tried out the Meta AI app and found that it's publicly sharing people's queries. 'Meta does not indicate to users what their privacy settings are as they post, or where they are even posting to. So, if you log into Meta AI with Instagram, and your Instagram account is public, then so too are your searches about how to meet 'big booty women,'' she writes. iPad for work: iPadOS 26 will bring new features to the 15-year-old device that might actually make it usable for a full day of work. A wave of recent headlines and posts has raised questions about Bluesky, from concerns about slowing growth to claims that the platform is turning into a left-leaning echo chamber and that its users are too serious. While those critiques capture part of the conversation, they don't reflect the full picture of what Bluesky is working toward. But if left unchecked, those perceptions could pose a real challenge to the platform's future growth. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
TechCrunch
2 hours ago
- TechCrunch
Week in Review: WWDC 2025 recap
Welcome back to Week in Review! We have lots for you this week, including what came out of WWDC 2025; The Browser Company's AI browser; OpenAI's partnership with Mattel; and updates to your iPad. Have a great weekend! The Apple experience: We kicked the week off with WWDC 2025, Apple's Worldwide Developers Conference, where the company showed off a newly designed iOS 26, new features across its products, and much more. There was considerable pressure on Apple this year to build on its promises and to make amends to developers as it lags behind in AI and faces continued legal challenges over its App Store. Snack hack: U.S. grocery distribution giant United Natural Foods (UNFI) was hit by a cyberattack, the company confirmed Tuesday. Much of UNFI's external-facing systems were offline, including web systems used by suppliers and customers, as well as the company's VPN products. Whole Foods was one of the victims, and it told staff that the cyberattack was affecting UNFI's 'ability to select and ship products from their warehouses' and that this will 'impact our normal delivery schedules and product availability.' Public debut: Chime's much-anticipated public debut finally arrived, with the company raising $864 million in its IPO. Iconiq was one of Chime's many backers taking a victory lap at its graduation to become a public company. This is TechCrunch's Week in Review, where we recap the week's biggest news. Want this delivered as a newsletter to your inbox every Saturday? Sign up here. News Image Credits:Google Not to be outdone: Google rolled out Android 16 to Pixel phones, adding group chat to RCS, AI-powered edit suggestions to Google Photos, and support for corporate badges in Google Wallet. Cabs are here: Elon Musk has spent years claiming that Teslas would be able to drive themselves. Apparently the time has come — maybe? Musk said this week that Tesla will start offering public rides in driverless vehicles in Austin, Texas, on June 22. Techcrunch event Save $200+ on your TechCrunch All Stage pass Build smarter. Scale faster. Connect deeper. Join visionaries from Precursor Ventures, NEA, Index Ventures, Underscore VC, and beyond for a day packed with strategies, workshops, and meaningful connections. Save $200+ on your TechCrunch All Stage pass Build smarter. Scale faster. Connect deeper. Join visionaries from Precursor Ventures, NEA, Index Ventures, Underscore VC, and beyond for a day packed with strategies, workshops, and meaningful connections. Boston, MA | REGISTER NOW An AI browser: The Browser Company said last year that it's going to stop supporting and developing its Arc browser, which, although popular, was never able to reach scale. The startup has since been busy developing an AI-first browser called Dia. And another one: OpenAI released o3-pro, which is a version of o3, a reasoning model that the startup launched earlier this year. As opposed to conventional AI models, reasoning models work through problems step by step, allowing them to perform more reliably in domains like physics, math, and coding. In other news, Sam Altman posted on X to say that his company's first open model in years will be delayed until later this summer. Desperately seeking: Now that people can ask a chatbot for answers — sometimes generated from news content taken without a publisher's knowledge — there's no need to click on Google's blue links. And that's hurting publishers. Cool? Mattel and OpenAI are teaming up to create an 'AI-powered product,' whatever that is. As part of the deal, Mattel employees will also get access to OpenAI tools like ChatGPT Enterprise to 'enhance product development and creative ideation.' 'A privacy disaster': Reporter Amanda Silberling tried out the Meta AI app and found that it's publicly sharing people's queries. 'Meta does not indicate to users what their privacy settings are as they post, or where they are even posting to. So, if you log into Meta AI with Instagram, and your Instagram account is public, then so too are your searches about how to meet 'big booty women,'' she writes. iPad for work: iPadOS 26 will bring new features to the 15-year-old device that might actually make it usable for a full day of work. Analysis A wave of recent headlines and posts has raised questions about Bluesky, from concerns about slowing growth to claims that the platform is turning into a left-leaning echo chamber and that its users are too serious. While those critiques capture part of the conversation, they don't reflect the full picture of what Bluesky is working toward. But if left unchecked, those perceptions could pose a real challenge to the platform's future growth.
