Hi everyone,
This edition’s article is about something I keep hearing from people who’ve started paying for an AI tool: they hit the usage limit and assume they need a bigger plan. Most of the time the fix is a clearer ask. I wrote up what I’ve seen, with the numbers behind it: Hitting your AI usage limit? Start with the prompt, not the plan.
The short version: decide what the finished thing should look like before you ask for it, put the whole job in one prompt, and leave the effort setting alone unless the problem is hard.
Worth reading
- Simon Willison, I built a new feature for my blog using my voice. He talked to a coding tool while cooking dinner, ums and all, and shipped a working feature because he knew exactly what he wanted built.
- Kevin Indig and Amanda Johnson, Managing the AI mandate upward (Growth Memo). Written for marketing teams, useful for anyone being told to “use more AI.” Their best line: “Tokens are the small cost. Your team’s hours is the big one.”
- Ethan Mollick, Management as AI superpower. Why people who are good at delegating to other people tend to be good at using AI.
- Latent Space, Reality checks on AI news. What happened when Databricks gave a coding agent to 3,500 engineers, and why they ended up giving it its own budget.
Around Springfield
The lunch at the efactory last week was great! Thanks to Pitt Technology Group and Springfield Tech Council for hosting and sponsoring these discussions. It was a packed house and definitely worth the price of admission.
By the way, this is only the second newsletter and if you’re still reading this far…thanks! I appreciate you coming along on the journey.
That’s it for this one. If a colleague is wrestling with the same questions, feel free to forward it.
Steve
Leave a Reply