In the third episode of the Platformer podcast, Claude Code creator Boris Cherny explains how he's automating his own job. He hasn't written a line of code in more than six months, and thinks the title "software engineer" could start to disappear as soon as this year. Platformer's Casey Newton talks to the inventor of the fastest-growing AI coding tool in the world about what Cherny actually means when he says coding is "solved," why he predicts companies will need both far fewer engineers and far more of them, what he tells new computer science grads about where to build a career now, and whether all this added productivity will ever let anyone work less. (Hint: so far, it just means doing more). Cherny also explains the surprise that keeps upending his own predictions: the people getting the most out of AI tools are increasingly electricians, doctors, and carpenters rather than professional engineers. Plus, Platformer fellow Ella Markianos joins at the top of the show to discuss a new Microsoft study on AI and jobs. Disclosure: Casey's fiancé works at Anthropic. Who else should we have on this show? Let us know at [email protected]. Sponsored by Atlassian Rovo | Become an AI-native team with Rovo. https://www.atlassian.com/rovo 0:00 Intro 6:23 Token Maxxing 10:52 Introducing Claude Code's Boris Cherny 17:21 Will software engineers disappear? 23:16 The Tractor Analogy 29:18 Is coding "solved?" 37:03 Productivity Paradox 42:20 Claude Cowork 48:56 Who's responsible for displacement? 52:55 Power users & the AI Divide 54:46 Predictions for next year
Comments 4
Sign in to join the conversation
Sign in
So yeah, Casey outed himself as a communist. But seriously on the topic, i don't think the reward mechanisms at companies move fast enough to congratulate employees for using AI. You've got once, maybe twice yearly review cycles where bonuses "could potentially based on a market conditions and financial goals" to then add everything else to the pile to try and reward, Like finance, seems like an easier thing for the executives and leaders to prosper from, but everyone downstream ends up suffering intstead of profiting
These “tool” analogies are just wrong. Calculators are tools, but AI is a tool user.
51:16 "We're an AI safety lab, it's weird that we even build product." Yeah, so, so weird!
Sorry but until you can provably replicate human ingenuity, this argument is just a waste of time. It's like saying the invention of the calculator was the end of mathematicians. The better the software engineer, the more complicated their harness setup is, and most of those setups do not look the same. To get quality code output, you have to know what you need the LLM to do, and more importantly, you need to know what the code itself needs to do. Having access to all of the ingredients and utensils doesn't make you a chef.