One idea changed how I'm thinking about AI workflows.

🔥

I used to think the best model would always give the best result.

After watching how Fable is being used, I realized that's only part of the equation.

The model isn't the real advantage. The process is.

Instead of asking the smartest model to do everything, let it design the approach, improve your prompts, and define the verification steps. Then let cheaper models handle most of the execution.

That idea really clicked for me.

The more I build automations, the more I realize good systems outlast any single model. Models will change. Prices will change. But if you own the workflow, the reasoning process, and the way your agents work together, you don't have to start over every time a new model launches.

I'm starting to spend more time improving my workflows than chasing the latest model.

Anyone else thinking more about process than model lately?

9 comments

One idea changed how I'm thinking about AI workflows.