Really excited to investigate hybrid search. I'd feel a lot more confident providing a list of links instead of a generated answer. Seen any good services that could help us out, here?
I don't think I mean to indict RAG + chat in general! I think it's totally possible that, if we put more work in, we'd get a great bot out.
But the bar is so, so high though. It's gotta be a truly great bot for us to not be scared of misleading our new users. And I'm still worried that "truly great" is going to take a LOT of work.
And for now, that's the problem. We're still a startup with limited resources. This tool isn't ready for us because we don't have the bandwidth to put the work in.
I can't wait til that bar drops, though. GPT 4o is a really solid step in that direction.
That much I will concede. I said we’ve had good results, but we’ve still been a bit scared to roll it out, more for potential cost and polish reasons than baseline quality, but of course I’m still worried about it saying something wrong.
Oh yeah, and I was worried about the "supervised/unsupervised" comment you made.
I'm not talking about supervised training. I think I mean to say that the OUTPUT is supervised/unsupervised. Like, I'm an experienced programmer, so I can supervise the output of Copilot, unlike our unexperienced docs users.
That's on me for not making that train of thought clear enough, and unfortunately choosing a term that's already in use by the AI/ML industry.
No yeah. You're totally right. I also see copilot lacking context, missing nuance, and frequently giving correct-sounding answers that actually aren't great.
I still think copilot is great for me, an experienced programmer who can recognize the bad.
But it is a bit scary that it's also empowering less-experienced end-user programmers to write bad code! It's the same story as our docs bot. AI is a good assistant, but it's not ready for taking over, yet.
That line was a fun, sassy one. And probably a bit too strong. I definitely meant to temper it with the next few sentences.
> Sure, it’s simple, which is worth a lot! … And for frequently changing, highly interactive pages like a dashboard, it’s probably enough
I spent years working on a medical dashboard SPA and we never once seriously talked about SSR. They totally have a time and a place for web apps. I agree with you and I’m glad you brought it up.
It’s a fair question. I think it’s a question I was hoping to address in this article, especially with the incremental migration pattern. If you already have a Next.js site, it’s honestly not too bad. If you’re not on Next.js, well, that’s a big migration either way. But considering server components during that migration might not be unreasonable — you’re writing React components either way, albeit RSC is a more complex mental model.
More specifically though: The docs migration took me about a month-and-change of work. Some of that was moving all of our styling to Tailwind. Some of it was (optional) RSC code gymnastics that I mentioned in the advanced patterns section, to see how many kB we could leave on the server. And much of it was the information architecture changes I talked about in that earlier blog post.
Meanwhile, marketing took us three months, but that number is complicated too, since we were rebranding the site, touching every component and page anyways. Most paths had like, 2 minutes of direct RSC work. Moving the calls to the data layer from outside our components to inside.
IMO not bad. In both cases, we had to work on the whole site, anyways. Might as well creep the scope a bit and move to the new React thing, since it looks like that’s where the puck is going.
