I imagined it like this
- 1.
I make a request to Buddy
- 2.
Buddy tries to figure out what I want and decides my request involves doing a, b, c on card c1, c2, c3
- 3.
Buddy tell me it made changes to 3 cards, shows the thumbnail. Little "show more detail" link expands when clicked
- 4.
c1 updated a, b; c2 updated a, c, c3 updated a, b, c.
- 5.
I uncheck c1 and c2, I only wanted one card updated; or I uncheck b that change was not intended
Something to that effect, not knowing if and how the prompt is broken down and that information is even stored or if it's a black box to gamma itself... then that would make anything like this quite an undertaking... 🤷♀️ I'm just here to share what I'm noticing as a heavy user of Agent as how it affected both my ability to trust the outputs on a deck where my collaborators had strong opinions about every little detail being right (and were inclined to not use Agent if it might make the wrong change and that change is hard to spot). But hoping there's a solution for this as I think it would go a long way with users.
to your point, I would not expect this level of control for a prompt that is in my history. For that I think restoring an older version of the deck would be appropriate, undoing individual changes from the past that might conflict with more recent history seems unreasonable to me. But who knows I'm sure there are other users out there more unreasonable than me 😂