Brainstorming the redesign with the team

I am redesigning my personal website with CodeYam. The current live site is pretty simple and straightforward, and I previously built it using v0. I have generated design mocks that I am not super happy with yet, so I want Dani's feedback from a design perspective and Jared's on how I can get better at prompting for the changes I want. There are six different variants, all with different layouts, designs and copy.

One thing I noticed while riffing is that certain fonts feel very commonly used by AI, and I went back and forth with Claude a little because there are certain defaults. I was trying to get something that felt more custom and original. One variant feels straight up professional but very dry, another feels a little more fun, and having something that feels beautifully designed is exciting to me.

Dani's advice is to create a Frankenstein of all the things you like from each one. On a personal website she likes seeing a picture of the person. Which one to choose really depends on who you want this to be for. Do you want to present yourself as a CEO and founder first, or as someone who works remotely and can work from anywhere?

Figure out the hierarchy of what you want people to see. Is the site about CodeYam, which is your current project, or is it a CV of the work you have done before, or is it more about you? Think about who you want to see the website first, whether that is customers or VCs, and lay out the sections based on that.

Jared's approach is to go through each one and rate it, and say what you like or don't like about it. You don't have to be specific, and saying you hate all of it is information too. Then ask the AI to be crazy for a few and explore ideas that have nothing to do with what you proposed. If eight mocks are generated at a time, take two or three to follow feedback, two or three as variations of your favourite, and two or three to be wacky so you discover something new.

Communicate your emotions, because at the end of the day it is going to feel right or not. Your attention gravitates towards something and that something is either good or bad, so describe that as best you can. As soon as I saw this I saw the photo and did not like the photo, or the font, or the colours, or the layout. Capture the first impression rather than trying to be technical about it.

In the era of AI, most of what it delivers is well designed by default, so taste is more subtle than whether something is well designed. Dani's answer is to follow your intuition and trust that what you like will work, and to gather a lot of references rather than starting from a blank page. Jared's is that calling it taste is a little off, because it is really about having more ideas about how to achieve something you like, and that is something AI can help you explore.

One thing I have done in this vein is a little silly. Rather than pointing at specific websites, I ask it to pretend it works at Pixar, or to be a designer in the Steve Jobs heyday at Apple, and approach my personal website that way. It gives a bit of direction while leaving it open enough to get genuinely different outcomes. Dani adds another version of the same trick, which is to put the AI in the designer's shoes and ask what questions it would ask you before building the site.