Hi everyone,
I recently came across git-ai and it looks like a promising approach to tracking AI-generated code, especially around compliance and IP considerations.
Before diving in on my own, I wanted to ask:
I’m particularly interested in how teams are integrating it into existing workflows and whether there are any pitfalls to watch out for early on.
Thanks in advance for any insights or pointers.
I wouldn’t expect this to work out of the box. In Gerrit, the primary identity is the Change-Id rather than the individual commit. Multiple commits sharing the same Change-Id are treated as patch sets of a single change, so review data (such as comments or notes) is associated with the change and evolves across patch sets.
Because of that, any migration would need to account for this model. Notes shouldn’t just follow individual commits, but rather be mapped across patch sets that belong to the same Change-Id.
I wouldn’t expect this to work out of the box. In Gerrit, the primary identity is the Change-Id rather than the individual commit. Multiple commits sharing the same Change-Id are treated as patch sets of a single change, so review data (such as comments or notes) is associated with the change and evolves across patch sets.
Because of that, any migration would need to account for this model. Notes shouldn’t just follow individual commits, but rather be mapped across patch sets that belong to the same Change-Id.
On Wednesday, May 6, 2026 at 12:32:54 PM UTC+2 Dezső Biczó wrote:Hi everyone,
I recently came across git-ai and it looks like a promising approach to tracking AI-generated code, especially around compliance and IP considerations.
> if they'll switch to use something like https://git-meta.com/
I have learn something new from this conversation! \o/ However, I assume this abstraction has the same mismatch with Gerrit's change ID concept like git notes (and therefore git-ai).