[crowcroft]

This is actually really exciting for Kagi. In a lot of situations the underlying model (Claude, GPT4 etc.) isn't that exciting, it's the connection to search to retrieve and summarize recent information that's exciting.

By already having a traditional search engine this puts Kagi at a big advantage compared to someone like Perplexity, or even Claude and OpenAI who I think are all cobbling together solutions on top of Bing's API.

[throwup238]

The ability to use lenses with custom assistants is the killer feature IMO.

Want to search for open source projects that implement some algorithm? Create a Github/Gitlab lens. Want to ask it questions only about some framework? Add it’s domain to a lens.

[crowcroft]

Yea, it solves a lot of limitations with LLMs generally struggling with 'niche' topics.

Being able to innovate on the search side while everyone else loses a lot of many training LLMs feels like a good space to be in at the moment.

[pjerem]

It’s interesting, I feel like LLMs are precisely good with niche topics where search engines are becoming absolutely garbage.

But maybe my search topics aren’t niche enough.

[crowcroft]

I guess it depends on what we define as niche. I'm thinking for coding type problems as an example smaller less well known languages or frameworks that don't have extensive coverage across multiple websites etc tend not to show up well in AI Chat.

If you can focus the chatbot on a specific docs website with a lens though that problem might go away.