After the new update, all actions I tell agent to do is slower and the output is not accurate. I feel like using replit is like magic before but now the magic is gone.
Any suggestions? or feel like the same way?
After the new update, all actions I tell agent to do is slower and the output is not accurate. I feel like using replit is like magic before but now the magic is gone.
Any suggestions? or feel like the same way?
I’m hoping improvements are on the way. Growing pains? Hang in there.
I feel that too. Even a 5 year-old kid could do better.
The Plan, Build and Advisor features we saw implemented last week on Agent 2 were a really good improvement. I was just getting the hang of how to ask the Advisor for deeper analysis and to use the plan feature to build task lists.
Agent 3 feels like a huge step backward. You can still plan and build task lists, but it feels exponentially more expensive and gets sidetracked and deviates from fixing issues way easier than it did before.
I’ve barely been able to do any development work on Replit today because Agent 3 is too buggy and unreliable.