---
title: "Comment by Mihai_LeanZero on \"Demystifying the AI Harness: What Makes Enterprise AI Actually Work\" | Hashnode"
description: "Agent = Model + Harness is a good way to frame it, and the RBAC and sandboxing points are the right ones to lead with. One thing I'd push back on a little is th"
url: "https://hashnode.com/posts/demystifying-the-ai-harness/6ab00ce83c33ac6bb081e8f4/comment/6ab04b4d74f20d749b6edd5d"
---

[Discussion](https://hashnode.com/posts/demystifying-the-ai-harness/6ab00ce83c33ac6bb081e8f4)

Comment [Demystifying the AI Harness: What Makes Enterprise AI Actually Work](https://hashnode.com/posts/demystifying-the-ai-harness/6ab00ce83c33ac6bb081e8f4)

[Mihai\_LeanZero](https://hashnode.com/@mihaileanzero)

[Sep 20](https://hashnode.com/posts/demystifying-the-ai-harness/6ab00ce83c33ac6bb081e8f4/comment/6ab04b4d74f20d749b6edd5d)

Agent = Model + Harness is a good way to frame it, and the RBAC and sandboxing points are the right ones to lead with. One thing I'd push back on a little is the model-swap claim -- in practice, swapping the model behind a harness isn't quite unplug and replug, the tool-call schemas and prompt structure that worked well for one model often need retuning for the next, especially anything relying on a specific function-calling format. The harness abstracts the plumbing but not the model's particular quirks.

[Akshay Gupta](https://hashnode.com/@akshay-gupta)

[Sep 21](https://hashnode.com/posts/demystifying-the-ai-harness/6ab00ce83c33ac6bb081e8f4/comment/6ab04b4d74f20d749b6edd5d#reply-6ab15a36a68226f49dd6a14e)

Yes, I agree. Swapping models is not plug-and-play. But the important thing is that the harness provides a abstraction layer and hides the model from end-users.