The AI sprint is hurtling toward a world where anyone can build personal concierges to do absolutely anything
Call your agents. Or better yet, code them—using sentences as dead-simple as this one. AI assistants that can handle work and everyday personal tasks, all powered by brisk English- language commands that require zero coding knowledge, are rapidly defining phase two of the AI boom.
AI tools like Anthropic’s Claude Code, Cursor and OpenAI’s Codex can now write and debug software, unlocking huge new sources of revenue. That success is pushing their makers toward a bigger ambition: automating our entire lives.
“ChatGPT started as something that could answer questions, but the long-term vision has always been a super-assistant,” said Nick Turley, head of OpenAI’s ChatGPT, one “that can actually help you get things done.”
What began as a way to autocomplete code quickly evolved into semiautonomous AI bots, or “agents,” that can work for hours on end with little human oversight. We can tell a bot to create a presentation for work, coordinate the family’s schedules and pick a March Madness bracket, all while it learns our personal preferences, no coding needed.
“For me, coding is kind of the new literacy, but luckily, it’s much easier to learn to code now than it was to learn to read, because you don’t have to practice, the tool just does this for you,” said Boris Cherny, who leads Claude Code at Anthropic. “But I don’t want to sugarcoat it. It is going to be very disruptive.”
The shift has permanently changed the lives of coders and sparked a $1 trillion market selloff as investors and executives contemplate the technology’s potential to reshape industries, including finance, legal and healthcare. Tens of thousands of job cuts have already been attributed to AI.
For OpenAI and Anthropic, winning the market for non-coders is the next frontier, especially as both companies race toward initial public offerings that could come as soon as later this year.
“When you think about the future of knowledge work, this is a multi-trillion dollar opportunity for companies,” said Denise Dresser, OpenAI’s chief revenue officer and former CEO of Slack. “It’s almost that if youcan think it, or you can describe what you want, you can build it.”
In Silicon Valley, AI agents have already become a way of life.
Venture capitalist Tomasz Tunguz uses them to create charts, blog posts and presentations. At one point, his spending on AI
reached $100,000 a year. That went toward subscriptions to AI tools from Google, Anthropic and OpenAI, as well as fees to access their models directly through application programming interfaces, or APIs.
“I’ll book travel, I’ll research a vacation, I’ll read the newspaper, all my email goes through it, my grocery shopping goes through it, music recommendations—I’m not reading magazines anymore—any question I have, everything I want to know about now goes through an AI,” Tunguz said.
Other users talk of building a dashboard to keep track of their child’s baseball stats, or automating the process of getting kids signed up for camps and daycare. The target market for these tools is simply “anyone who needs to do work on their computer,” said Felix Rieseberg, the engineering lead for Cowork, Anthropic’s new feature designed for nontechnical tasks.
Rieseberg became a first-time dad in January and has used Cowork to keep track of and analyze a large folder of medical records. He’s also used it to apply for a mortgage and compile expense reports.
Tunguz, founder of Theory Ventures, estimates that agents could generate $36 billion in annualized consumer revenue in the near future, an amount that would represent a stunning growth rate from virtually nothing just over a year ago. But he says that’s just the start—the real money will come from lucrative enterprise contracts, a much larger market opportunity than chatbots.
Anthropic and OpenAI charge about $200 monthly for the most expensive tier of their AI tools. In February, Anthropic said that Claude Code was generating $2.5 billion in annualized revenue. OpenAI declined to disclose Codex revenue.
Like many power users, Tunguz uses the tools to orchestrate the work of many agents simultaneously.
“If I use a chatbot, I am having a one-on-one conversation with an AI, but if I use something like Cowork, I can have 15 to 50 simultaneous conversations with AI and that means they sell me a lot more inference,” said Tunguz, referring to the AI processing power required to run these systems. “That means more business for these companies.”
Convincing millions of new users to adopt agents will require overcoming a lot of concerns. People are losing sleep over the idea that using these tools means inadvertently training a model to replace them, or that they represent the beginning of widespread job loss and an eventual economic collapse.
There are concerns about the security and safety of these systems, too, as instances of violence and self-harm linked to chatbots become more common and stories circulate about agents going rogue and deleting files. Letting AI complete tasks for you often means giving it free rein over your data. And it can feel like babysitting a fleet of agents that are constantly messing up.
The coding wars, which set the stage for phase two of the AI boom, have been under way for several years behind the scenes, with a host of AI companies competing to win over software developers and influential early adopters.
One of the most important of those companies is Cursor, a smaller startup that played a huge role in kick-starting the coding battle when it launched a tool in 2023 that made it faster and easier to code. Cursor works with most AI models, meaning a user can tap Anthropic’s one day and OpenAI’s another day, or other models being produced all over the world.
Cursor became a Silicon Valley darling by ushering in what insiders call the “vibe-coding” era—basically allowing non-engineers to build software, apps or websites without knowing how to write a line of C or Python. Nvidia CEO Jensen Huang has called it his “favorite” AI tool.
Last valued at $29.3 billion, Cursor has grown to about 400 people and recently passed $2 billion in annualized revenue, doubling in a three-month period. It’s still growing despite a recent barrage of X comments and blog posts declaring it dead, as people wondered about the rise of what tech insiders are dubbing “Cursor killers” like Claude Code and Codex.
The reason that expectation persists is because both OpenAI and Anthropic have been subsidizing the cost of usage on their platforms. They’re charging far less than it costs to run the AI, much like Uber and Lyft once did when they offered rides for just a few dollars as they raced to push each other out of the market.
Doing work with an AI agent, especially one that can work for hours without stopping, can be very costly. Cursor charges through subscription tiers with usage limits. To lure customers, Anthropic and OpenAI have been offering power users far higher token limits, like $1000worth inside a $200-a-month plan.
Cursor has expanded from a focus on coding to more broadly “helping developers build the factory that creates their software,” Michael Truell, the startup’s young, redhaired leader, recently wrote on X.
A spokesperson for Cursor declined to comment.
Claude Code was released in early 2025 and developers sang its praises, claiming it was more capable and reliable than Cursor, and able to run autonomous agents instead of serving as a coding co-pilot that required constant human input.
The tool went viral in November when, just before Thanksgiving, Anthropic released an updated model of Claude that made the tool even more powerful. In December, it spread beyond software engineers and began growing at an exponential rate.
Some users said they were “Claude-pilled” or on “Claude benders.” They sported hats embroidered with the product’s crab mascot and stopped Cherny on the street for selfies or autographs. Suddenly, developers weren’t writing any code, they were managing several agents simultaneously.
Demand for OpenAI’s Codex exploded this winter when the company introduced its latest model, which has far superior coding abilities than its predecessors. Over the last 30 days, OpenAI has witnessed a “fundamental switch” in how businesses are using AI, said Dresser, OpenAI’s revenue chief, who called it a “wildfire moment.”
Codex has more than two million weekly active users, OpenAI said. The company is finalizing a strategy to shift focus toward coding and business users, scaling back side projects that have unnecessarily pulled focus, the Journal has reported.
Traffic has increased eightfold over roughly the last two months, requiring it to constantly bring more computing power online.
To keep top users happy, AI companies are treating them like VIPs, responding swiftly to their complaints on social media, or even wining and dining them.
When OpenAI was debating how to make its model’s personality more “fun,” Codex product lead Alexander Embiricos spotted a power user arguing on X that the model should actually be more terse and concise. The feedback helped the company decide to give users a choice of personality, he said.
Anthropic co-founder Ben Mann and others at the company have held events for power users. Over Chinese food in the Gangnam District in Seoul in December, coders shared ideas with Anthropic executives about how to make the agents more reliable.
AI startup worker Sigrid Jin, who attended the Seoul dinner, single- handedly used 25 billion of Claude Code tokens last year. At the time, usage limits were looser, allowing early enthusiasts to reach tens of billions of tokens at a very low cost.
Jin flew to San Francisco in February for Claude Code’s first birthday party, where attendees waited in line to compare notes with Cherny. The crowd included a practicing cardiologist from Belgium who had built an app to help patients navigate care, and a California lawyer who made a tool for automating building permit approvals using Claude Code.
