according to reports, anthropic engineer boris chorny stated that in his current programming workflow, thousands of ai agents automatically work for him overnight. chorny, the creator of claude code, shared his ai working model during an interview with sequoia capital on may 4.
anthropic’s claude code creator: thousands of ai agents automatically write code for me at night
the details he revealed during the interview—including that he mostly runs these ai agents via his smartphone—highlight a shift in how some engineers in silicon valley are using ai: they no longer treat ai systems as chatbots but rather as autonomous assistants operating online around the clock. demonstrating his phone to the audience, he said, “i’ve installed the claude app. once i open it, there’s a small code tab on the left, and i can simultaneously launch numerous task sessions.” when asked how many sessions he runs at once, chorny said he typically opens five to ten sessions, each containing multiple ai agents.
he added, “usually every night, i have thousands of ai agents engaged in intensive development work.” chorny noted that he heavily relies on two key features of claude code designed specifically for continuous automation: loops (/loops) and routines. he explained that users can set up loops locally using cron jobs, while routines execute periodic tasks on the server, allowing ai agents to keep running even when engineers close their laptops. “this is simply the coolest feature,” he said. “if you haven’t tried it yet, i highly recommend giving it a shot.”
this isn’t the first time chorny has shared his approach to using ai agents. back in january this year, he publicly posted this agent configuration on the social platform x, describing it as “actually quite conventional and straightforward in design.” during the interview, chorny admitted, “it’s kind of funny—i didn’t realize this would surprise anyone. for me, it’s just a standard part of my daily programming routine.”
