There’s a new flex in Silicon Valley: It’s not “how big is your headcount,” it’s “how big is your agent swarm?”
In the last six months there have been big leaps in agentic capabilities
for artificial intelligence. These programs take a large language model out of the chatbot and set it loose to actually do stuff in the world, like book a flight or delete your junk emails.
Coding agents like Anthropic’s Claude Code or OpenAI’s Codex focus on building software. Then there’s OpenClaw, and its imitators, which let you run agents by messaging them from your phone. They can work autonomously, 24/7, on whatever goal you give them. Though caveat emptor: as with any generative AI, the results can be a bit unpredictable.
Still, you’d think having an army of AI minions might free up some time and make work more chill. Well in Silicon Valley, not so much.
John Huang, a tech industry veteran who runs his own startup investment network, spent the last few months trying to automate as much busy work as he can. “I have, right now, eight bots running,” Huang said. “So for example, in the morning, at 6 a.m. it goes to Reddit, looks at the top threads and the topics I'm interested in and I told it, ‘By 7 a.m. in my inbox I want a report on the top news items I'm interested in.’”
He has a bot updating his resume and another checking his competitors websites for changes.
Huang used to pay for a human assistant in the Philippines. Now he just pays for tokens — the unit of measurement for AI workloads.
Last month, he hosted a meetup of the agent-curious at a Menlo Park Starbucks, down the road from all the big venture capital firms. A crowd of about a dozen showed up: entrepreneurs, Big Tech workers and hobbyist tinkerers.
“I feel like our whole company should just shut down and focus on this for about a month,” said Kevin Staight, who runs sales strategy for
Zime, an AI startup. At the time, Staight was mostly using agents to organize his emails and write research reports. But he was working on building a custom networking assistant to track who he meets, map connections and prompt him to follow up.
“If I adopt this now I probably have a three- to four-month runway before everyone else catches up,” he said. “And so I want to be ahead of the curve to be more efficient.”
A sense that you’re falling behind if you’re not 10xing productivity, even while you sleep, has taken over the industry, said Nikunj Kothari, a venture capital investor at FPV Ventures in San Francisco.
“Everybody has this feeling of like, ‘Hey, time is the only thing that matters. And in that given unit of time, which we don't get back, how can I have AI do a lot more for me than the next person?’” Kothari said.
He calls the phenomenon “ token anxiety.” And he sees signs everywhere in San Francisco — people keeping tabs on their agents during parties, at bars, even while outside touching grass.
“I would see, like, laptops slightly open, and like the warm glow of the light,” while walking through Dolores Park in the Mission district, he said.
Kothari himself is running agents for email, market research and data analysis. He’s replaced his evening Netflix time with Claude Code, dreaming up new tasks to automate just for fun.
“I have two young kids at home, and I felt guilty because I'd be like, ‘Oh, go to bed quickly so I can get back to my computer,’” he said. | 
