In the coming months, hundreds of thousands of brand-new coworkers will be entering the workforce. You’ll learn their names and recognize their profile pictures. They’ll get added to your email chains and Slack channels. They’ll become chiefs of staff, engineers, and marketing gurus.

But these coworkers aren’t human: They are all artificial intelligence agents.

The rise of AI agents—systems, based on large language models, designed to complete tasks autonomously—has launched a tsunami of companies selling the services of digital employees. These agents are supposed to be the perfect worker bees: available 24/7, never complaining, and providing superhuman intelligence.

Yet there has been curiously little discussion of what these agents are like as actual coworkers. Julie Bedard, a Boston Consulting Group partner, noticed the murky nature of how people talked about AI at work three years ago, as rapid adoption of ChatGPT convinced executives the technology could bring exponential improvements in productivity. “People didn't have language for it,” Bedard tells WIRED. “Is it a tool? Is it a teammate? Is it a colleague? Is it a coworker?” When platforms such as Microsoft’s Copilot and Google’s Gemini for Workspace launched in 2023, they were marketed primarily as tools for employees. But in January, when Bedard and colleagues polled 1,261 managers, 22 percent said their organizations had added AI agents to corporate org charts. Bedard said that this percentage is likely to grow as dozens of startups offer up the services of AI employees and big tech companies hype their autonomous agents. In June, for example, Microsoft launched “Scout,” an agent built to complete tasks like rescheduling meetings and drafting emails.

“Your company essentially hires your assistant,” Omar Shahine, the corporate vice president of Microsoft Scout, said when the agent launched. “The whole point of having a personal assistant is that they're working when you're not working.” The success of AI agents depends not only on their ability to provide financial returns, but also to work harmoniously alongside the existing workforce. That means companies have crafted agents that boast the benefits of AI, packaged in a human-like bundle of traits to encourage employees to anthropomorphize their new colleagues.

“Your company essentially hires your assistant,” Omar Shahine, the corporate vice president of Microsoft Scout, said when the agent launched. “The whole point of having a personal assistant is that they're working when you're not working.” The success of AI agents depends not only on their ability to provide financial returns, but also to work harmoniously alongside the existing workforce. That means companies have crafted agents that boast the benefits of AI, packaged in a human-like bundle of traits to encourage employees to anthropomorphize their new colleagues. “People start to really respect the agent and form strong attachments," says Dhruv Amin, CEO of Anything, a vibe coding startup that entered the AI coworker market in August with a platform called Skydive. “They'll talk about it like humans.” Skydive’s agents “aren’t chatbots or workflows,” but coworkers, with their own names, roles, and cloud computers. They work across Slack, email, iMessage, and more. Each is represented by a Muppet-esque avatar, which Amin described as “close to human, but definitely not human.” They have names often related to their roles; Canary, for instance, monitors production 24/7.

“We think it's important to personify them a little bit,” Amin says, “because most people still don't know what a real agent is.” Christine Wendell, CEO of Pronto Housing and an early Skydive adopter, uses agents for a laundry list of busy work: writing contracts, scheduling meetings, sending follow-ups emails to customers. Her agents are, essentially, perfect employees.

But how are they as coworkers? Skydive agents not only embed in every form of workplace communication, they continually evolve to match coworkers’ communication styles, requests, and corrections. In other words, they become better and better at fitting in with their actual coworkers. When Pronto’s engineering team began using a coding agent, Wendell noticed that employees started saying that they “worked with Alice”—the agent—on a project.

Yet part of the agents’ appeal is that they are not human. Beyond their limitless work ethic, they also exist outside of office politics or professional neuroses. When “Bob,” the agent serving as Wendell’s chief of staff, errantly listed all the details of Wendell’s calendar in a Slack channel, she didn’t feel the need to sugarcoat her swift correction.

“I'm definitely very aware that it's not human,” Wendell says. “I dump on them in a way I wouldn't do to a human, because it would not be polite.” AI coworkers are entering the job market just as agents are increasingly infiltrating our personal lives. While Silicon Valley has been using agents such as Open Claw and Instinct for months to complete annoying tasks, Meta’s newly launched Muse aims to push personal AI agents into the mainstream.

Muse, like Skydive’s muppets, is represented by a cute and fuzzy little avatar. In the words of WIRED senior writer Reece Roger, the user experience is something “like texting a friend with task requests—the AI agent acknowledges messages with a thumbs-up emoji and gets to work in the background.” By using existing relationships as a blueprint, these agents—whether they’re AI coworkers or, like Muse, “built for everyone”—offer a shortcut to understanding how to use them.

Anthromorphizing AI comes with specific risks in a work context. Research from Bedard and her colleagues at BCG found that, when managers were told work was completed by an AI employee, they caught 18 percent fewer errors than when told it came from an AI tool. Agents also don’t necessarily fit in the same boxes as human employees, Bedard says, in their abilities or their weaknesses.

“Companies are just starting to wrap their heads around what guardrails” are even applicable, Bedard says.

People forming deep emotional connections with AI is nothing new; people have been dating chatbot companions for years at this point. Even in the workplace, AI tools have sparked some curious reactions. Last year, Henry Blodget—the founder of my former employer, Business Insider—wrote a blog about building a team of digital subordinates, and having a “human response” to the headshot of a particularly attractive AI colleague.

“After a few decades in the human workplace, I’ve learned that sharing certain human thoughts at work is almost always a bad idea,” Blodget wrote. “But did the same rules apply to AI colleagues and native-AI workplaces?” Sarah Franklin, the CEO of HR company Lattice, has more existential concerns. Lattice added digital employees to the org chart two years ago, before the concept had fully entered the mainstream. Franklin emphasizes that Lattice uses this model to encourage accountability—not to encourage workers to treat AI colleagues as if they were human.

Franklin sees such anthropomorphizing as a strategy by tech companies to get users hooked on their services, unable to resist the dopamine trigger of an emotional connection. While this can help companies build an active user base, Franklin believes it may have detrimental neurological and behavioral repercussions for the human workforce.

“We have these models that have consumed all human knowledge, and are so proud to have these multitrillion valuations—but are potentially destroying humanity,” Franklin says. “It is very scary. And what is the purpose? Is the purpose to put people out of work?”