It would be nice if there were different "modes" for Kagi Assistant, such as a "Basic" and "Advanced" mode, that impacts how configurable the user-selectable models are. Currently, Assistant operates in what I would call an "Advanced" mode where the user is left to sort through many different models and has to decide what's best for their use case very specifically. It would be nice if there was a Basic mode where the user was presented with fewer choices and the specific model was completely abstracted away from the user. Instead, it would be nice if Kagi maintained a preferred list of "Basic" models and instead of classifying them by name for the user, they classified them by use-case, with only one model per use-case.
For example, in Basic mode, the model drop down may offer the user options like:
- General Questions
- Coding
- Problem Solving
- Research
- Image/Media Analysis
...and that's it. To the "Basic" user, they could very quickly decide which of these activities they want to do with the AI, and they could just trust the Kagi team's judgment as to which model would be best for that activity. A user could always switch to "Advanced" mode and go back to choose-your-own adventure with AI models if they want something very specific or more tailored to their use-case, but I think that the majority of people probably would appreciate a "Basic" mode most of the time.
This would enhance the model selection currently available in Assistant. A user would switch between "Basic" and "Advanced" model selection modes based on their level of comfort with the various AI models as well as their estimate of the cost/benefit of playing around with different models for their needs.
