@kurissuuu For me the platform, and the business, of the company I work for has always used AI. So in a way, I could say I've been working in an AI-native business for the past 6+ years I guess. Not too precious about it, TBH.
Shipped an internal tool for our content design team today.
We send health related messages across a lot of channels and languages. The team had no way to interactively see what those messages actually looked like in context. I went looking for something off the shelf. Found nothing. So I built it, off and on, over a couple of weeks.
The idea is simple. Add base English content, let it propagate down into the various channels. Channels other than email come with their own constraints, so you can edit channel and language specific variants from there.
The part I didn't expect: the team is already learning things they didn't know about their own content. Truncation, localization quirks, brand variation. Stuff that was invisible before now shows up the moment you preview it.
Couldn't find it on the market. So I made it.
@tomjohndesign Had a somewhat related experience where the sales team wanted a UI to look better on a projector. We ended up buying a few different ones and setting up a war room to adjust the UI until we achieved the best compromise. Good times.
Used the latest OpenAI image model to generate different poses from a photo. Wanted to make the standard designer hero section feel a little more alive. Seems interesting at the very least.
This image is what the model produced after a ton of tweaking and prompts. Quality and accuracy degrade pretty quickly. Ended up having to constantly provide source material as reference.
Once it felt good enough to use as a base, I took it into Photoshop to make it look like the other illustrative images I'm using on my site.
Anyway, you can see it in action at: https://t.co/jFbpdIJtLO
Not exactly how I want it yet, but close.