Ahnii!
Casey Newton interviewed Boris Cherny, the creator of Claude Code, for Platformer last week. Most of the coverage pulled the headline-friendly quote that coding is "solved" and moved on. I want to point at a smaller moment in the same interview that I think matters more. This post is about that moment, and three things I've watched up close that line up with it.
Newton asks Cherny about the AI divide. The worry, in shorthand, is that the people who already have power will use these tools to get more of it. The early data on who benefits from new technology usually goes that way. So who's actually getting the most out of Claude Code?
Cherny's answer surprised Newton, and it surprised me when I read it. He talks about a recent Anthropic hackathon for the Opus 4.7 release, and says the people who won were largely not professional engineers. "There was an electrician, a doctor, a carpenter who used it to build an app." Same pattern at the 4.6 hackathon. He calls it a continuous surprise. The people who get the most value out of Claude Code, he says, are not the people he'd expect.
That's a small line in a long interview. It's the line I haven't stopped thinking about.
The AI-jobs discourse is mostly fighting over one question: will engineers be replaced. Cherny's answer to that question, if you read the whole interview, is actually mild. He thinks the title "software engineer" probably changes ("builder" is the word he uses), the role expands, and the number of people writing code with the help of agents goes up roughly a hundred-fold. That's not an extinction story. It's a transformation story, with a hiring pull, not a layoff cliff.
But that's not the only question we should be asking. The other question is the one Newton was getting at: when the syntax gate falls, who walks through?
The hackathon detail is an early answer. Not "the engineers got faster." An electrician, a doctor, and a carpenter built apps that won. The default story we tell about new technology — the people in the room get more powerful first — doesn't fit. The people in the room didn't even win.
Here's what I'm seeing in three different places.
I just finished my Grade 12 coursework as an adult learner through Sagamok Anishnawbek's Lifelong Learning Centre. AI was in the loop for a lot of it. I'm not the demographic anyone meant when they said "AI for developers." I'm a forty-something who came back to finish his OSSD, and the tools worked for me anyway. I finished the work that for years had been hard to finish. That's one shape.
I started a petition asking the Rainbow District School Board to publish a real public AI policy before next semester. Students across the board are using these tools every day. The adults setting the rules haven't caught up. Partway through my own coursework, I was pulled aside and questioned about my AI use because there was no policy to point at. The petition is about that, but it's also about the kids: they're walking through a door before anyone has written a rule about it. That's another shape.
I'm building Anokii, the embedded chat on the OIATC site, on top of my own framework. Anokii isn't being built for engineers. It's being built so a member of a community can find a community-specific resource and get an answer that cites where it came from. Per-community variants. A relevance gate so it stays quiet when it shouldn't speak. A topic-confidence gate on the citations. When the tool is reachable, and built with the people who'll use it in mind, the people who reach for it are not the obvious ones. That's a third shape.
Three different angles. An adult learner finishing high school. A petition asking for a fair rule. A chat that surfaces community resources without pretending to be smarter than it is.
None of these are "engineers replaced." All of them are "someone got access to a tool that wasn't available before, and the question is whether the systems around them caught up."
Cherny's hackathon detail isn't an outlier. It's the leading edge of what happens everywhere when the syntax-and-credentials gate falls. People who were never in the room start showing up. The electrician builds the app. The adult learner finishes the coursework. The community member finds the resource they needed and gets a real citation instead of an authoritative-sounding guess.
The energy we spend arguing about whether engineers will exist in five years is energy we are not spending on what the hackathon detail is actually telling us.
Three places I'd want it to go.
Schools need policy now, not after the next semester. The Rainbow District petition is one example. Every board with students using these tools is facing the same problem. The students are already there. The policy can't be retroactive forever.
Communities deserve tools built for them, not demos pointed at them. Anokii is the version of that I'm running, and the way to tell whether a community tool is real is whether it cites its sources, whether it knows when to stay quiet, and whether it gets better when the community pushes back on it. Demos do not pass any of those tests.
For engineers and builders and whatever else we get called next: stop pretending the threat is your job disappearing. The threat, if there is one, is that the tools become most useful to the people who already had power. Cherny's data so far says that is not what's happening. Don't make it happen. (If you want the engineer-facing half of this argument, I wrote the playbook for taking a vibe-coded prototype to something a stranger can run earlier this week. This post is the other half.)
The people getting the most out of these tools right now are not the people the discourse expects. The electrician winning the hackathon, the adult learner finishing the diploma, the community member finding the resource. None of them are the story we're telling. All of them are the story we should be telling.
That should change how we talk about this.
Baamaapii