Replace Chef AI's "Ask Sablier" with "Ask in ChatGPT"

[smol-ninja]

Query	Chef AI	ChatGPT

Its evident that ChatGPT (or any other popular LLM) is better at responding compared to Chef AI. Ask in ChatGPT button can be added to every page (e.g. Foundry and Pimlico) redirecting users to the ChatGPT interface with a prefilled prompt.

Please research and analyze this page: https://docs.sablier.com/concepts/use-cases so I can ask you questions about it.
Once you have read it, prompt me with any questions I have. Do not post content from the page in your response. Any of
my follow up questions must reference the site I gave you.

ChatGPT does a good job analyzing all the pages even when you just give it https://docs.sablier.com/ instead of the specific page. The response in the above screenshot was generated by referring to the docs website and not any specific page.

Thoughts @maxdesalle and @PaulRBerg?

[PaulRBerg]

I think there is some value in having an iframed AI bot in the docs site.

However, the value isn't high enough to justify paying for Chef AI, and I'm sure they will eventually want to charge us.

So I'm in favor of this.

[smol-ninja]

I think there is some value in having an iframed AI bot in the docs site.

I completely agree with this. But I also think soon all popular docs generator will provide some kind of integration with popular LLMs (its quite obvious at this point and then tools like Chef AI will become redundant and outdated).

I looked up into the Docusaurus Github and found this PR which will have "Ask AI" integration into their default DocSearch + Algolia that we already have (powered by an LLM of our choice which is cool).

If we can fill this form, they might put us on the Beta. Its asking for Algolia application ID which I don't have.

[PaulRBerg]

Great finding with that Docusaurus PR.

Re Algolia app ID - you can find it in docusaurus.config.ts.

[smol-ninja]

Great. I have filled the form.

[smol-ninja]

Not required after #342. We replaced Chef AI with Algolia Ask-AI search.

[NatanTechofNY]

@smol-ninja Awesome for you to integrate Ask AI so quickly. If you are looking for higher accuracy answers, I would suggest you read our new documentation on Markdown indexing through the Crawler: https://docsearch.algolia.com/docs/v4/askai-markdown-indexing

This is a manual workaround, we will be releasing an automated way to do this in the next week or so. However, if you are fine with the answers you are receiving, feel free to disregard this solution!

[smol-ninja]

Hi @NatanTechofNY, good to see you here. We are definitely considering the markdown approach (see #336).

we will be releasing an automated way to do this in the next week or so

That'd be awesome. How can we be notified once its released?

if you are fine with the answers you are receiving

They are Okish right now, could definitely be better.

[NatanTechofNY]

@smol-ninja I can definitely let you know here once it's out for the Crawler, in the meantime if you're comfortable with adjusting the Crawler configuration yourself, feel free to follow the docs. We also post updates in our Discord every now and then