#build-with-ai
#feedback
#questions
Quick architecture question.
I’m setting up a long-term workspace for a client. She’s an author/expert but completely non-technical.
I need one canonical workspace that both of us can access from different computers, while agents can also work against the same files and everything stays synchronized/versioned.
My underlying architecture is human-readable files have to remain the source of truth. Any memory/retrieval layer can sit on top, but I don’t want the project locked inside a proprietary system.
The cleanest option I’ve found so far is private GitHub + local Codex, possibly Codespaces for her so she doesn’t have to deal with Git. I’ve also looked at a shared VPS/Tailscale layer if agents eventually need a persistent environment.
Have you solved something like this before? What architecture would you use today if one collaborator is technical and the other one is non-technical?
I’m setting up a long-term workspace for a client. She’s an author/expert but completely non-technical.
I need one canonical workspace that both of us can access from different computers, while agents can also work against the same files and everything stays synchronized/versioned.
My underlying architecture is human-readable files have to remain the source of truth. Any memory/retrieval layer can sit on top, but I don’t want the project locked inside a proprietary system.
The cleanest option I’ve found so far is private GitHub + local Codex, possibly Codespaces for her so she doesn’t have to deal with Git. I’ve also looked at a shared VPS/Tailscale layer if agents eventually need a persistent environment.
Have you solved something like this before? What architecture would you use today if one collaborator is technical and the other one is non-technical?
3 comments
Sign in to join the discussion.
if you want, you can turn this into a web app, and implement claude agent SDK inside it.
you can both access the app, and work on it, with the same claude agentic capabilities