points by p1necone 2 days ago | 0 comments
I've been kinda disappointed with the $10 plan tier of the AI assistant. It looks good at first glance - shows you that it's doing a bunch of research, you can see which sources it's looking at etc. But then when it goes to report results on anything data heavy it makes a lot of mistakes with putting figures under the wrong headers, using incorrect values etc.
The last time I used it heavily was a few months ago so it might have gotten better, and I assume the more expensive tier where you can choose which models it uses might be a bit better depending on which model you choose. I much prefer just pointing my own local agent harness at a search api at this point though.
But since I'm on the Ultimate plan, for stuff that matters I usually use Fable with a custom assistant (prompt), and sometimes the stock "Research" assistant. Both work really great for me. I use the "Research" assistant for more technical questions, it tends to dig up information from forums and cross-reference it with the source-code file it digs up from github and quote the RFC that it has directly fetched too.
Though I usually prefer my custom assistant prompt, which Kagi seems to combine with their general prompt in the background. This works really, freaking great most of the time. My custom prompt is an elaborate version of "answer the question, but then try to second-guess what the user is doing and also provide a different perspective that could be more useful than a direct answer". This has derailed my narrow train of thoughts a few times and turned me around into the right direction.
That said, I haven't played around with my own local agents for search yet to compare (in part because Kagi does a very good job at this), and this is something I want to try, in a "own your tools" sense. But the first thing I'll make sure that it uses the Kagi API for its search.