Your Prompt Is the Product: Working with o3 & o4-mini-high in the Agentic Era

A hands-on playbook for turning OpenAI's newest reasoning models into agents that think, verify, and act—plus key prompt simplifiers that work with more advanced, tool-using reasoning models!

Great products change our view of what’s possible, and o3 has definitely done that for me in the last week or so. (Yes, it’s only been a week). This post is about taking some of the learnigns I’ve been stewing on and getting them out there for you all to learn from, dig into, and try for yourselves.

As I wrote, I have this line rattling around in my head: the line from technology <> product <> magical customer value has never been so direct as it is with AI. That immediate product resonance is why ChatGPT is on track for 1B users by end of year, just about 3 years after launch. If you’re counting, Facebook took nearly 9 years to hit the same milestone.

Why call out customer value in an article about prompting? Because new models are valuable when they make this line to customer value clearer and simpler, and when they do so they demand new prompts. That’s it.

And o3 and o4-mini-high both do that. And they deserve their own prompting guide, which is really an update from my pocket guide to prompting.