I've ended up using a few models for different things. One tends to write better, another reasons through problems better, another I just use for quick stuff. On paper that's great, in practice I spend a stupid amount of time getting each one up to speed
Every time I switch I'm basically re-explaining the same background. Here's the project, here's what we already figured out, here's the docs that matter. The conversation in any single model is fine, it's the constant re-briefing across all of them that eats my time
And it's not just pasting text. Each one remembers a slightly different version of the project depending on what I told it last, so I'll get answers that contradict each other because one model is working off context the other one never got
I've tried keeping a master doc I paste in everywhere, but I forget to update it, and then I'm back to square one
How people who run multiple models actually handle this. Do you keep one external source of truth and feed it into all of them? Pick one main model and only use the others for one-off tasks? Or just accept that context lives in silos and move on?
[link] [comments]