Learn how to build GPT-3 AI apps with OpenAI and @Vercel edge functions. In this video, I explain how I built and how you can build your own GPT-3 powered apps:

I'm excited for this video because it presents a solution to an interesting serverless timeout issue for folks on Vercel's hobby tier. In the video, I explain edge functions from scratch and run a benchmark to compare them to serverless functions both in terms of speed and UX.

If you'd like to watch with timestamps and in 4K, feel free to check this same video out on YouTube!

@OpenAI @vercel 🤯🤯🤯 wow streaming is a gamechanger, thanks for sharing Hassan! Gonna try this out in my future AI projects 😄

@OpenAI @vercel My pleasure dude! Can't wait to try out your replicate webhook solution too 🫡

@OpenAI @vercel The streaming API is a very powerful tool, thanks for showing it off! However, I would not generate the prompt on the client as this can lead to people exploiting your API with unwanted prompts.

@OpenAI @vercel My pleasure! That's a good point – I'll probably add a check on the server to make sure it only accepts the prompt that I defined on the client.

@OpenAI @vercel Wow that difference between streaming/not was night and day! Epic video explainer!

@OpenAI @vercel Thanks so much Daniel – means a lot coming from you!

@OpenAI @vercel You say that one can't use @prisma in Edge Functions, but you might want to clarify that Vercel Edge Functions on supports the HTTP protocol in Edge Functions and therefore for Prisma you'd need to use their Data Proxy: