Just walk away, it's not worth it.

Wer in der Union hat sich eigentlich gedacht, dass es eine gute Idee sei, Kai Wegner als Bürgermeisterkandidat aufzustellen und den auch noch zu wählen?

Cookie-Warnungen als Bankrotterklärung der EU-Bürokratie, als Monument der Unfähigkeit und des Scheiterns. Ganz wunderbarer Rant vom Autor des Web-Frameworks Ruby on Rails

In other news: I’m writing again, working title “Building for the new user”

This video by LangChain explains the fundamentals of context engineering in just 14 minutes. Context Engineering is effectively the #1 job of engineers building AI agents. do not sleep on it.

As a product designer, here are 8 of my favourite designers on X: (with a breakdown of their best shots) 🧵 1. rauno - show me a more interactive website, I'll wait

Prism 1.0 is coming soon! I just want to revisit streaming text chunks and get that dialed in.

Static text embeddings is growing on me after combining it with hybrid search and query expansion(s) for write intensive workloads. CPU friendly (just a embedding table lookup) and infinite context length.

Also, what always happens with CIA/FBI crimes and misconduct is happening now: Those who point it out when it's happening are called crazy because (most of) the evidence is hidden. When the proof is finally revealed, we're told: why focus on this? It was so long ago!

Here's the workflow: - Ingest documents and generate binary embeddings. - Create a binary vector index and store embeddings in the vector DB. - Retrieve top-k similar documents to the user's query. - LLM generates a response based on additional context. Let's implement this!

2️⃣ Generate Binary Embeddings Next, we generate text embeddings (in float32) and convert them to binary vectors, resulting in a 32x reduction in memory and storage. This is called binary quantization. Check this implementation 👇

Claude code demonstrated the power of agentic search. It started a movement away from fuzzy code embeddings, vector databases and single-turn RAG.

Turn PDF files into clean, LLM-ready data! Dolphin is an open source document parsing framework that converts PDFs into structured formats like Markdown, HTML, LaTeX, and JSON. 100% Open Source

Apple’s Embedding Atlas (MIT license) lets you explore, cluster, search, and filter millions of points in real time with WebGPU. Great for AI, data science, and research. Demo: apple.github .io/embedding-atlas Source: github .com/apple/embedding-atlas

Boost just released. 🚀 Ashley Hindle calls it your AI coding starter kit for Laravel. No hunting for MCP servers or cobbling together rules. Everything’s built in so you get higher quality AI output from day one. Hear more in our latest episode.

ℹ️ Der Krieg begann nicht erst am 7. Oktober. Immer wieder argumentiert die Propaganda, der Gazakrieg habe nicht am 7. Oktober begonnen. Mit diesem manipulierenden Framing soll im Kopf des Empfängers ein Bild entstehen. Das impliziert, das Jahrhundertverbrechen sei irgendwie

The application of AI to marketing is transformative. Counterintuitively, many teams are devoting too much effort to it: - Implementations of AI tend to focus exclusively on the outputs of these tools. Most often this means creative production. But much of this functionality

Product development has completely changed for us. Three steps. One seamless flow. Step 1: Lovable. Sketch. Prototype. MVP. Ideas take shape fast - design to build, all in one place. Step 2: GitHub + Copilot. Delegate repeat work. Track issues, review PRs. No manual coding -

JSON prompting for LLMs, clearly explained:

If you try to build everything your customers ask for, you won’t build a great product.