We know that AI doesn't always get it right... That's why we prompt it, train it and test it But sometimes when it goes wrong you just have to laugh Here are two examples from my image creator using Google's Nano Banana. The first is obvious - it chopped my head off The second is a little more subtle and it took me a moment to realise what the error was. Can you spot it? 🤣

I'm building a Speed to Lead WhatsApp workflow bot for a Salon client and I've reached that point.... Too many hair, nail and massage conversations! Every conversation could lead to a tweak in the prompt.. Which requires another set of tests.. Which leads to more tweaks.. And more tests.. BUT! This is how we get to a point of feeling comfortable that our bots will do the majority of what we want them to do. And have the knowledge that we can still make minor changes when needed. How much testing do you do before you set something live? P.S This one still hasn't gone live yet as they want to change some wording/style so guess what - there will be more tests required!! 🤣