Coding with AI can feel intimidating to those who haven’t tried it, but there’s a useful analogy for those trying out coding agents for the first time.
Anthropic researcher Boris Cherny says that an AI model like Claude is the horse, and a coding assistant like Claude Code is the harness. In the analogy, the human is still in the loop, using the harness to direct the horse to where they want to go.
“People have tried to make AI do coding for the longest time and to automate more and more of coding in various ways,’ Cherny said on a podcast. “And it hasn’t really worked. I think probably for a couple of reasons. One is the models weren’t really good enough and the second one is that the scaffolding, the thing on top of the model, wasn’t good enough,” he added.
“And when we initially launched Claude Code, it like, sort of worked. I remember when we launched it, we gave it to the core team and we noticed that engineers were already using it in a couple of days. It was just the craziest thing. The model wasn’t very good, the harness wasn’t very good, but even in this early version, it was already a little bit useful. And I think that over the last year what’s happened is the model has gotten way better at agentic coding. And obviously the harness is Claude Code, because the way you interact with the model is you can’t just like directly use it, you have to use a harness. It’s sort of like if you’re riding a horse, you need some sort of saddle, and that saddle makes a giant difference when you’re riding a horse,” he explained.
Claude Code is a terminal-based AI coding assistant developed by Anthropic, allowing developers to delegate coding tasks, analyze codebases, and interact with development tools like Git using natural language commands. It integrates into the existing development workflow without requiring separate applications, understanding the project’s context and acting as an agent to execute tasks like code migrations and bug fixes. Claude Code was launched only in February 2025, and is already generating more than $500 million in run-rate revenue, and its usage has grown more than 10x in the last three months.
And what this analogy also serves to emphasize is that the human programmer is still in the loop. Without a human, a horse wouldn’t know where to go, with or without a harness, but with an expert human rider, it can be steered to a specific destination. And that might be the state of coding for a while — humans with a horse will be much faster than those on foot, and humans who are experts at riding horses will leave everyone else far behind as they use AI models and tools like Claude Code to supercharge their workflows.