Recently, I’ve seen quite a lot of users complaining about their experience with the Agent model, and although some of the suggestions and frustrations make sense, I just wanted to show through this post that you can get your desired output by fine-tuning your prompts. Now, it does consume a lot more credits than getting the exact output you wish for, but trust me, to some extent we need to adjust to what’s being provided and make the best of it for the time being.
In the link attached, you’ll find a chat that demonstrates how I developed my latest portfolio for an application to Vercel for the role of a Community Engineer. I wanted to make an unforgettable portfolio that revitalizes its viewers, the person doing their job of scouting resumes should also have fun once in a while…
You can see I faced many challenges at several steps where I kept asking it to make something but it wouldn’t hit the nail exactly where I wanted it, so I modified my requirement in a manner that both v0 Agent and I are satisfied with the outcome 
Sorry, I was unable to attach a video and too lazy to create an online link, so you can see the kind of interface I wanted to greet the user at landing:
But if you visit the link attached below, you’ll find something that some of you may find to be better than what I shared here 
Portfolio link: https://rizas-portfolio.vercel.app
Chat: https://v0.app/chat/rizas-portfolio-nDm7y3KUwNn
So, please try your best and guide the Agent to achieve the kind of output that you can be satisfied with, at least for an MVP.
P.S. All this rant aside, I do agree with the users requesting for the model selection option we had earlier while keeping the Agent as one of the options, I couldn’t find a logical reason as to why it isn’t possible tbh
Thank you for