I’ve been having some amazing successes with agentic AI and coding lately. A run-down post on these would be interesting. You can find one write-up at Goodbye Wordpress, thanks AI. With this as the backdrop, let me ask you a question.
Should you amp up your AI at the start of a project?
We’ve heard the debates about whether you should be polite and thank you AI. I think yes but never as a separate follow up comment. But should we try to get it HYPED and excited to work on the project?
I just published my first solo book. One of the challenges is that on the buy page there is a simple pair of images/buttons: Buy on Gumroad and Buy on Amazon. Things are rarely as simple as they seem. The Kindle version of the book is available in 12 different locales (US, Canada, Germany, etc.). Each one of these is a different URL based on the location of the web visitor! That means the website should adapt to each visitor and point them at their store if possible. To accomplish this, I combined some magic from GeoIpLite, diskcache, and a text file full of links to various Amazon stores’ listings of my book. Rather than grinding through this, I asked Cursor and Claude Sonnet 4.5 to help make that page dynamic.
I am finding huge success if I have a top-tier model work with me to create a detailed, reviewed plan. Then have the AI work step by step through the plan.
I was excited for the book and wanted my coding buddy to share in my excitement!
And it delivered! “I’m absolutely ready to rock! 🎸”
I’m really enjoying this. I think going forward, each time I kick off one of these detailed, plan-based projects, I am going to hype my AI coding buddy. Even if it doesn’t make a difference, it makes it more fun. :)
BTW, I’m working on an Agentic Coding course at Talk Python. Join the mailing list if that sounds interesting. There will be plenty of excitement there too.
Cheers
Michael