👀 New video out! Check out my top five tips for better AI-assisted using ChatGPT or Claude Sonnet! #ai

@huntabyte @bitworks_studio @pavel_stianko Wow, this looks like a million bucks! Fresh!

🌱 Just released Svelte Render Scan - a visual debugging tool that helps you track component re-renders in your app. Useful for debugging reactivity and performance issues in your Svelte apps! Based on the excellent render-scan library by @nullvoxpopuli

@nullvoxpopuli @elonmusk Where I live you can submit your taxes with a text message on your phone. You don't even need the internet. The US system has always seemed bad and not moving in the right direction unfortunately.

RT @joyofcodedev: Here's how signals work on a basic level to help you understand the magic behind them. Signals hold a value and notify s…

@Sikandar_Bhide @arvidkahl 👋 You can run the distilled R1 versions locally in Ollama (or full one if you have like 400GB RAM). Longer context also eats memory, we'd need to experiment with how much RAM is needed. If you want to experiment you can use R1 full on with 128k context

@wolfr_2 2025 version of the turbo button

@dummdidumm_ it all makes sense now 😅

@uwukko > company with 2.5T market cap

@Mhamad_6000 They are similar, but Render-scan works for any framework.

@wongmjane Because it shows that the OpenAI 157B valuation is wildly exaggerated. If VC valuations were normal I think people wouldn't complain as much.

@budivoogt @SourcegraphCody @cursor_ai 👋 I'm using the same workflow with Claude Projects that I showed in the video below. With Projects I can add my entire codebase into the LLM, and the performance is amazing. Doing that via API is very expensive. Sending partial code doesn't work as well.

@SvelteSociety This is 🔥🔥🔥

This is the quintessential AWS experience. I just want to call an LLM using AWS Bedrock - 175 permissions to configure.

Evan is so right. Frameworks and libraries should be so simple anyone can use them. Anything else is poor DX design.

@GroveEsten @FredKSchott @youyuxi In this case, it's both Reacts poor baseline perf combined with Next.js worst-in-class SSR performance.