Coinbase CEO Brian Armstrong fired engineers who refused to use AI assistants

In a decisive move, Coinbase CEO Brian Armstrong revealed that he has let go of engineers who resisted the adoption of AI coding assistants. This announcement came after the company introduced enterprise licenses for GitHub Copilot and Cursor, tools designed to enhance coding efficiency. During a recent appearance on the Cheeky Pint podcast, hosted by Stripe co-founder John Collison, Armstrong expressed his surprise at the slow uptake of these new tools among staff. "I went rogue," he admitted, explaining that he was taken aback when informed that some employees were hesitant to embrace the AI solutions despite the company's commitment to provide access. In a bid to accelerate adoption, Armstrong took to Coinbase's main engineering Slack channel, issuing a directive requiring all engineers to register for the AI assistants by week's end. "AI is important. We need you to all learn it and at least onboard. You don’t have to use it every day yet until we do some training, but at least onboard by the end of the week," he stated. Armstrong further emphasized the importance of this initiative by scheduling a follow-up meeting for those who failed to comply. At this meeting, Armstrong noted that while some engineers presented legitimate delays, such as being on vacation, others could not justify their lack of participation. "I jumped on this call on Saturday, and there were a couple of people that had not done it. Some of them had a good reason, but others didn’t. And they got fired," he revealed. Acknowledging the heavy-handed nature of his actions, Armstrong recognized that not every employee agreed with his approach. Nevertheless, he maintained that the decision was necessary to signal the critical importance of AI integration within the company. Since then, Coinbase has ramped up its focus on AI, instituting monthly sessions where teams can showcase innovative applications of AI technologies. As the conversation shifted, Collison raised concerns about the potential pitfalls of relying too heavily on AI-generated code. Armstrong concurred, stating, "I agree," underscoring the need for careful consideration in managing an AI-driven codebase.