Tagged “instructions”
-
Making the next switch cheaper
This one was expensive because everything lived inside a product. Four files kept outside any account make the next switch an afternoon.
-
Rebuilding custom instructions somewhere else
Custom instructions are the highest-value thing you cannot export. Rewriting them as a plain-text file you own means you never rebuild them again.
-
What does not transfer
The export covers what was said. Everything that made the assistant feel tuned to you lives outside it, and has to be rebuilt by hand on the other side.
-
Seeding a new assistant deliberately
A new account starts cold. Loading your instructions, reference documents and a short context brief on day one is faster than letting it learn again.
-
Copying memory out before you clear it
What the assistant remembers about you accumulated over months and cannot be regenerated. Read it, copy it, sort it — then clear.
-
Rebuilding projects and folders on the other side
The organisational layer has no export format anywhere. Rebuild only the containers you used, from your own list rather than by copying.
-
The reference files you re-upload constantly
A style guide, a spec, a sheet you attach every week. Uploads live inside the account you are closing; the canonical copies have to live outside it.
-
Building a prompt library you own
The prompts you use weekly are stored inside a product you are leaving. Written as one plain-text file, they stop being something you rebuild every switch.