3 Likes
“Launching today when the team wakes up” -Amjad
3 Likes
It’s live now! https://x.com/pirroh/status/1894434712623747294
Yup, I’ve been using it for the last 20 minutes and so far I love it.
1 Like
NGL the results so far are pretty good. What model is it using?
I don’t know for sure, but if I had to guess the agent uses Claude 3.7 sonnet thinking. Just a guess.
Does anyone know if it now handles replit object-storage? Close agent 1, claude 3.5, wud butcher my entire code base every time i wanted file handling changes.
When will we get claude 3.7 in assistant? I need an assistant that understands replit object-storage SDK, and doesn’t completely rewrite my entire code base to fs every time i want to change file handling
1 Like
I think it is using Claude 3.7 Sonnet because I have noticed a marked improvement in the way it gives me feedback and walks me through the code since I activated “Explorer” in my account.
It may be a little too eager in some respects, creating tests upon tests while we were building a process. We ended up with 20 files that ended up getting whittled down to 5 we actually used in our process. So there’s that to watch out for.
Getting to the end result was much easier and I feel like an automation I’ve been working on for weeks now (off and on) is finally in a stable state. I’m not a developer by trade, but I’ve been doing UX and product design for a long time so I have a good grasp on describing processes and writing product definition docs. This tool has been so empowering!
Excited to continue building with this new, more powerful Agent! 
Sorry to say, v2 is terrible with the replit object storage, as was v1.
I’m really impressed so far with the performance of agent v2! Making lots more progress and breaking through roadblocks with this new agent and model! Thanks, keep up the great work!