@derrickcchoi@jxnlco It’s great but still one step away from using the connectors that we need for work like slack, notion, etc. lag in response is fine while it does this work
@stolerlyn I get it but there are probably a few easier heuristics.
- nothing longer than a page (50 pages reports can be one page recommendations with everything as appendix for support)
- it shouldn’t take you less time to create it than it takes the review to consume it
- etc
@darrowoah Assume you mean coding as a the use case? Yeah. You shouldn’t use medium for most of that. It doesn’t give itself enough time to understand the dependencies and design choices
@blader I was doing this for a while with opus and 5.5. Do you find fable stands its ground on design choices? I find it caves to gpt on perfectly reasonable choices just because gpt is more opinionated
@jxnlco The step that I’m waiting for is when gpt live can work for enterprise and run the connectors to action. I don’t care if it doesn’t respond for one minute. Being able to read the slacks, write the notion, set up the meeting, etc. all through voice is going to be amazing.
@quintendf He does it to himself through competing through direct comparisons rather than pure differentiated product. Seen here again. Same with ‘Bloomberg killer’ et el