Anything, a San Francisco AI startup, runs 150 autonomous agents alongside 15 human employees. Founder Dhruv Amin believes this 10:1 ratio will become standard within a year. The agents handle scheduling, coding, SEO, onboarding, and office management — tasks that previously required human bodies and human salaries. The agents don't sleep, don't complain, and don't file HR grievances. That's the pitch, anyway. The adoption curve is steep. McKinsey reports 40% of large organizations used AI agents between May 2025 and June 2026, up from 27% the prior year. Boston Consulting Group found companies expect to shrink headcounts by 10% to 15% as delegation to AI scales. Lindy, which builds customizable agents, laid off a third of its staff and replaced open roles with bots. In the San Francisco startup scene, the agent-to-human ratio has become a status symbol — proof you can do more with less. The human experience of this shift is more complicated than the pitch deck suggests. Rafael López Muñoz at Wordware found that after deploying agents for recruiting, he spent most of his time fixing their mistakes — mixed-up candidates, hallucinated résumé details. Tarek Abillama at River described the universal transformation: "We all became managers." The skills that matter have inverted. Years of domain experience matter less; the ability to prompt, coordinate, and quality-check bots matters more. The customer-facing friction is real and growing. Brooke Hopkins at AI voice startup Coval has banned agents from customer communications, calling it lazy. Duncan McClure, an AI model evaluator, described being "hounded" by AI recruiting agents and concluded the message was clear: the employer didn't care enough to put in human effort. River's sales agent Liv gets flirted with by customers and pressured for off-script discounts — people anthropomorphize the bots even when told they're artificial. Jay Blumenfield's production company in Los Angeles offers the clearest window into the labor substitution at work. After firing an assistant director mid-shoot, he replaced the role with an agent that produced schedules, call sheets, and script materials within hours. Now he and his partner run the entire company with a suite of agents — no assistant, no location scouts, no social media staff. Work that took two to three people two to three weeks happens in hours. He acknowledges it feels "a little taboo" in Hollywood but says the economics left him no choice. The productivity paradox cuts through the hype. MIT research shows companies are not seeing major productivity gains from AI in final outputs, even as individual task speed increases. Amin himself admits he hasn't stopped working weekends despite his 150-agent army. The agents accelerate throughput but don't reduce the total work envelope — they expand it. You get more done, but you don't get done sooner. Lindy's Haneen Azhar captures the tension most honestly. She worries about her own role in eliminating jobs — agents now handle product announcements, social media, and marketing videos that roughly 100 people in Lindy's creator program used to produce. She worries about the next generation of workers: "If I was trying to get my foot in the door right now, I'd be worse off than I was a year ago." The people building these systems can see the downstream consequences. They're building them anyway.
