Mike rudge

@matt_elevenlabs 11B

We just raised $500M at an 11B valuation 🎉 To celebrate, we’re giving away 1,000 free credits so you can test our platform. For the next 6 hours, comment “11B” below and we’ll DM you the credits (must follow) 👇

@asidorenko_ Does this work with vscode / copilot?

"How can I use react-best-practices skills?" Codex example 👇

@aisdk I feel like the codemod caused more issues than solved. It got a lot of the server messages wrong.

We want to make AI SDK 5 migration smoother. If you upgraded, what issues did you hit? If you haven’t, what’s holding you back?

@haydenbleasel Nice! How does this work with hosting on vercel? Things like realtime transcriptions wouldn’t work. And timeouts are likely on long form audio?

The new `transcribe()` function works just like Orate’s transcription did, letting you plug in different provider models with the same simple interface. You point it at audio data and specify which transcription model to use, and it gives you back the result.

Last big update of the week: Orate is now part of the Vercel AI SDK! I recently joined the team at @Vercel, and one of the first things I've been working on is merging Orate’s speech synthesis, transcription, voice changing and audio isolation features into the AI SDK.

The debounce feature of @inngest is 🔥

@aisdk Is it possible to pass a reference image yet? “Style this room…” kinda prompt

Generate images with OpenAI's latest gpt-image-1 model.

@vercel_support you have just sent me a report for a project that is not mine. 😳

@rauchg Great stuff! Do you think you will be looking at a solution for handling realtime applications? I hate having to use other services to run another server to handle websockets for example

This brings the edge worker / runtime chapter for Vercel to a close. The lessons: ◆ Developers want full, open runtimes. A cloud runtime that promises efficiencies and has DX and ecosystems tradeoffs ain't it. We brought the compute efficiencies and cold start improvements to the existing ecosystem instead, and that's Fluid. ◆ "Edge" is for routing, static assets, and prerenders… not for apps, APIs & databases. Fluid is multi-region, optimizing for running your app next to your Postgres, as God through the laws of physics intended. But multi-region is not dozens or hundreds of CDN edges. Google Spanner, Turso, and Amazon DSQL are leading the charge here in helping you go global in an economical and actually performant way. ◆ It's over for v8 Isolates as a cloud primitive. If we can bring I/O efficiency to microVMs with broad runtime support, the huge security tradeoff in using *the JS engine* as your virtualization layer becomes… a strange value proposition to say the least 😵‍💫 In fact, as of last week, all Edge Middleware runs on Fluid in the Vercel cloud. As I've shared[1], it's actually even faster and more efficient than before (and runtime agnostic, with @nodejs coming soon to @nextjs), while also being more secure. [1] x.com/rauchg/status/…

Fluid (vercel.⁠com/fluid) tl;DR: ◆ Serverless DX, Servers efficiency ◆ Full @nodejs & Python runtimes (more soon) ◆ `waitUntil`, streaming, cold start prevention… ◆ Even faster and more efficient free hobby tier ◆ One click to enable & zero-config experience 👇

@upstash @togethercompute This is rough on mobile. I used your demo and it crashed my phone is so hot now and it’s all misaligned. Maybe not quite production ready yet.

Powered by @togethercompute and the Vercel AI SDK for easy integration into existing projects. The perfect starting point for adding a chat to your next project. 100% open source. 👉🏻 Live demo: rag-component.upstash.app 👉🏻 GitHub: rag-component.upstash.app

Introducing a beautifully designed AI chat component! 🎉 ◆ Ready to use, drop into any existing project ◆ Stream AI responses in real time ◆ Built-in RAG functionality Beautiful out of the box, fully customizable. 30-second demo 👇🏻

✨ Tailwind CSS v4.0 is here! Huge performance improvements, radically simplified setup experience, CSS-first configuration, modernized P3 color palette, container queries, 3D transforms, expanded gradient APIs, @​starting-style support… …and tons, tons more.

@wesbos I know you’ve talked about it a million times. Transcribing long form audio server side (2+ hours). Are you still using whisper API? (I’m on serverless as well so need a async solution)

@dakdevs @mattpocockuk Don’t think so. It’s just a wrapper around the underlying models. And they support it natively.

@Mike_rudge @mattpocockuk You’d likely need to use a 3rd party service in conjunction with a tool for that.

My first AI course is complete... And it's free! It's a 16-lesson tutorial on Vercel's AI SDK. It's the missing piece for working with AI in TypeScript. No bullshit, no hype. Just solid, practical advice.

@mattpocockuk Yeah weird ain’t omission from the sdk ain’t it. I wonder if it was to safe guard against abuse? I did wonder if you could base64 encode the image and send it with a prompt or something but pretty sure that it wouldn’t work.

@Mike_rudge Looks like no - generateImage only takes in a prompt, not a messages array.

@mattpocockuk Depending on your needs, Gemini can do a pretty decent job of this natively. Something like this

Anyone got a decent guide anywhere for transcribing audio with Vercel's AI SDK?

@huozhi Figuring out why a page is not static. Would love to dev tool to help understand why a page is rendering the way it is.

What’s the annoying errors you’re facing in Next.js that you have no clue how to fix it? Lmk

After 4 yrs of work: mmm.page 1.0 is live! Dead simple, drag-&-drop websites. Add text, images, shapes, GIFs, videos, code — even draw. It's your internet canvas.

@thomascarrollgp wins Insurance Broker of the Year! Sponsored by @AvivaUK #BIA2024