AI Agents Become Coworkers—and Managers of the Bots

By Sophie Lindqvist

About 40% of large organizations used AI agents in some capacity between May 2025 and June 2026, up from 27% a year earlier, according to McKinsey’s State of AI Global Survey. Boston Consulting Group expects companies to scale back their workforces by 10% to 15% over the next few years as they delegate work to AI.

MIT research shows that companies are not seeing major productivity gains from AI in the present moment. That gap matters because vendors are already selling semi-autonomous agents for sales, marketing, coding, recruiting and other core operations.

What does an AI-heavy workplace look like?

On an average workday, Dhruv Amin chats with dozens of colleagues at the AI startup Anything. He asks Gwen, his executive assistant, to schedule a meeting; Cody, his coding partner, to troubleshoot a technical problem; Betsy to handle search engine optimization; and Juno to onboard customers.

If snacks run out, he sends Justin, the office manager responsible for stocking the kitchen, a Slack message. None of these colleagues is human; they are AI agents, or software that carries out tasks without human intervention.

Anything runs with 15 human employees and roughly 150 agents. “I think in a year, every company will have more AI co-workers than human co-workers,” Amin said. At Anything, agents have become the org chart, not a tool sitting beside it.

Why are AI agents becoming a status symbol?

Several startup employees in San Francisco said agents have become a status symbol, showing that a company can do more with less. They described workers bragging about agents that answer emails, help with essential operations and give sales pitches, while agents alienate customers and colleagues who seek human interaction.

At Wordware, chief of staff Rafael López Muñoz said: “That’s the status symbol.” He described outsourcing work to agents as “an extreme time saver.”

Viktor sells a customizable “AI chief of staff,” while Lindy offers agents for sales, customer support and administrative work. OpenAI’s Dots, Anthropic’s Claude Cowork and Microsoft’s Scout can each integrate with thousands of other tools and handle projects after employees leave for the day.

Their pitch is an always-available coworker that never clocks out, complains or creates human resource problems. Startup employees also said delegation costs less than hiring, which makes a high agent-to-human ratio a source of pride. The business case is simple: more output per employee, with agents taking over work once assigned to people.

How many workers can one laptop replace?

Lindy once employed almost 50 people. After laying off one-third of its staff, the company says it has about 20 employees and has filled many open roles with AI agents, though it continues hiring for some positions.

Haneen Azhar, Lindy’s senior growth marketing manager, said: “People are constrained by having two hands and 24 hours in a day. Now, that constraint has gone away, because your one laptop screen can be divided into 16 different agents that are running in parallel. These tools make you feel so powerful.”

Azhar has outsourced three parts of her own work:

• Managing her calendar.
• Testing how customers respond to different marketing taglines.
• Generating brand images.

She also worries about her own role in eliminating jobs as she offloads work to agents. Lindy’s agents now handle product announcements, social media posts and marketing videos, work that about 100 people in its creator program used to handle.

“If I was trying to get my foot in the door right now, I’d be worse off than I was a year ago,” she said. At Lindy, lower headcount and broader agent coverage arrived together, but entry-level opportunity looks narrower.

Can agents cover sudden staffing gaps?

Jay Blumenfield runs a Los Angeles production company that makes vertical microdramas, short scripted series designed for viewing on a phone. His schedules and budgets are tight, so when he fired an assistant director on a recent shoot, he used an agent instead of making a last-minute hire.

Blumenfield uploaded the script and prompted the agent to do everything the assistant director would have done. It produced the schedule, call sheet and day’s script, then sent the materials to the cast and crew. Everyone got their work done.

“It would have been an unmitigated disaster, but this kind of saved us,” he said. Now, Blumenfield and his partner, Tony Marsh, run the company alone with a suite of agents. “We don’t even have an assistant anymore,” he said.

Their agents read scripts, run social media and plan travel. Blumenfield created an agent called Shirley to act as chief of staff, plus another that scouts and books locations for upcoming shoots. He normally would have had to hire two or three people for two or three weeks; agents now complete the work within hours.

He called the shift a little taboo, though filmmaking economics left him little choice. It helps the company afford microdramas starring actual people, at least for now. His warning reflected broader fears in Hollywood that AI could end creative work.

“The game in Hollywood is that you’re not allowed to talk about using AI. I don’t want it to replace people. But we should be honest about how it is helping us,” Blumenfield said. For Blumenfield, agents made a small production possible, while the human cost remained unresolved.

What changes when employees manage agents?

Some workers who rely on agents say the technology cannot replace the human workforce entirely. Even so, agents are changing the skills employers need and forcing some employees to become overnight managers of bots that often make mistakes.

Agents can write code, which reduces the value of routine programming skills. They still require humans to prompt them carefully, judge their output and provide substantial oversight.

At Wordware, López Muñoz said he let agents run the startup’s recruiting, then spent most of his time undoing their mistakes. They mixed up candidates with similar names and invented parts of résumés. “The qualities we’re looking for are changing,” Azhar said.

Azhar now prioritizes the ability to coordinate and delegate over years of experience. Tarek Abillama, founder of River, which makes an agent for sales calls, said workers must constantly check AI’s work and give it feedback. “We all became managers,” Abillama said.

“We have to be great at multitasking, because now you’re running six agents who are doing different tasks.” River’s first agent was a clone of Abillama that delivered sales demos. He later replaced it with Liv, a female persona that researches customers and their companies, demonstrates the product and handles the sales process in any language at any time of day.

Even though River tells customers that Liv is an AI agent, Abillama said people appear to treat her like a person. He has seen transcripts where customers flirt with her, and others have tried to push her off script and request bigger discounts. The new core skill is not typing the task; it is directing agents and catching their errors.

Will customers accept an automated coworker?

Brooke Hopkins, founder of San Francisco-based AI voice startup Coval, uses agents to speed up internal work but views them as tools rather than colleagues. She has barred her company from using agents to communicate with customers, calling the approach lazy and sending the wrong message. Hopkins said the loss of human interaction had become a point of tension.

“I can immediately look at an email and know whether or not it was written by AI,” she said. “If I get an email and it feels like a human, then I usually will read it.” Duncan McClure, a contractor who evaluates AI models, recently received repeated emails and phone calls from AI agents seeking to hire him.

“The message it sends is that we don’t care enough to actually put in any effort,” he said, adding that the outreach turned him off from working with those employers. Recruiters may save hours by using agents, “but it’s my time that they’re wasting.” Automation may save employers time while pushing customer trust onto workers who remain visible.

Does automation mean working less?

Several startup employees said that automating costly, time-consuming parts of a job does not necessarily mean working fewer hours. Amin, for example, gets more done with agents but still works on weekends.

The adoption figures show that agents are spreading faster than the research supports as a general productivity boost. The next test is whether employers can turn delegated work into measurable output without making every worker a permanent supervisor of bots.