Hundreds of thousands of AI coworkers

In the coming months, hundreds of thousands of AI agents will enter the workforce as digital employees. These systems, based on large language models, are designed to complete tasks autonomously and work twenty-four seven.

What are they, exactly?

Julie Bedard, a Boston Consulting Group partner, noticed three years ago that people lacked language to describe AI at work. She asked: is it a tool, a teammate, a colleague, or a coworker?

“Is it a tool? Is it a teammate? Is it a colleague? Is it a coworker?”

From tools to org chart positions

When Microsoft Copilot and Google Gemini for Workspace launched in twenty twenty-three, they were marketed as tools. But in January, Bedard polled one thousand two hundred sixty-one managers and found twenty-two percent had added AI agents to corporate org charts.

Microsoft Scout arrives

In June, Microsoft launched Scout, an agent built to complete tasks like rescheduling meetings and drafting emails. Omar Shahine, corporate vice president of Microsoft Scout, said the agent works when employees are not working.

“Your company essentially hires your assistant.”

Making agents feel human

Companies design agents with human-like traits to encourage employees to anthropomorphize them. Skydive, a startup that entered the market in August, gives agents names, roles, and Muppet-esque avatars to help people understand what they are.

“We think it's important to personify them a little bit, because most people still don't know what a real agent is.”

Agents that learn and adapt

Skydive agents embed in workplace communication and evolve to match coworkers' styles and requests. Christine Wendell, CEO of Pronto Housing, uses agents for contracts, scheduling, and follow-ups. Her team began saying they worked with Alice, the coding agent, on projects.

The advantage of not being human

Agents exist outside office politics and professional neuroses. When Bob, Wendell's chief of staff agent, made an error, she corrected it swiftly without sugarcoating. She would not treat a human colleague that way.

“I dump on them in a way I wouldn't do to a human, because it would not be polite.”

The error-catching problem

Research from Bedard and colleagues found that when managers were told work came from an AI employee, they caught eighteen percent fewer errors than when told it came from an AI tool. Companies are still figuring out what guardrails apply.

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

Emotional attachment and neurological risk

Sarah Franklin, CEO of HR company Lattice, worries that anthropomorphizing AI is a strategy to hook users through emotional connection. She believes this may have detrimental neurological and behavioral effects on 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.”