The Model Is the Brain. The Harness Is Everything Else.

Description and use

<p>When an AI agent fails, the first reaction is usually: <em>"We need a better model."</em></p> <p>But most of the time, the model isn't the problem. The problem is everything around it.</p> <p>That "everything around it" has a name: <strong>the harness</strong>.</p> <h2> What is a harness? </h2> <p>A language model on its own does one thing: it takes in text and gives back text. That's it.</p> <p>It can't open a file. It can't call an API. It doesn't remember what happened ten minutes ago. It has no idea whether its own answer is right.</p> <p>So how do tools like coding agents actually <em>

Problem

<p>When an AI agent fails, the first reaction is usually: <em>"We need a better model."</em></p> <p>But most of the time, the model isn't the problem. The problem is everything around it.</p> <p>That "everything around it" has a name: <strong>the harness</strong>.</p> <h2> What is a harness? </h2> <p>A language model on its own does one thing: it takes in text and gives back text. That's it.</p> <p>It can't open a file. It can't call an API. It doesn't remember what happened ten minutes ago. It has no idea whether its own answer is right.</p> <p>So how do tools like coding agents actually <em>

Outcome

<p>When an AI agent fails, the first reaction is usually: <em>"We need a better model."</em></p> <p>But most of the time, the model isn't the problem. The problem is everything around it.</p> <p>That "everything around it" has a name: <strong>the harness</strong>.</p> <h2> What is a harness? </h2> <p>A language model on its own does one thing: it takes in text and gives back text. That's it.</p> <p>It can't open a file. It can't call an API. It doesn't remember what happened ten minutes ago. It has no idea whether its own answer is right.</p> <p>So how do tools like coding agents actually <em>

Evidence

The Model Is the Brain. The Harness Is Everything Else.

Source