I’ll start with a confession. I use Fable, Claude’s biggest model, all the time, and more often than I’d like to admit I forget to switch it back. Which means the next thing I ask, something as trivial as where the nearest pizza place is, gets the same heavyweight thinking I’d want for a board paper. That is a very expensive way to find a margherita.
It’s Everyday AI Friday, and this week’s video is about a setting almost nobody has been shown: effort level.
Are you blowing through your AI usage every month and buying extra tokens to get to the end of it? Are the answers you get so thin they’re barely worth reading? Or do you ask a simple question, wait, wait some more, and wander off to make a coffee?
If any of that sounds familiar, the problem may not be the tool. It may simply be set to the wrong effort level, and it takes seconds to change.
In ChatGPT, it’s a dropdown right in the chat box, where you can move between instant, medium and high. I leave mine on medium unless I need something quick or something thorough, and when I’m genuinely researching a topic I go to the plus sign and choose deep research, which can take up to half an hour. In Claude, the dropdown sits in the same place but gives you more to work with, because you choose the model first and then the effort level, all the way from low to max.
If you lead a team that has recently been handed AI seats, this is worth five minutes at your next team meeting. Most people are working on whatever setting happened to be there the day they first logged in, and nobody told them it was a dial. You see it in the quality of the work, and you see it in the bill.
What’s yours set to right now? And has anyone in your organization ever actually been shown where to change it?
