Hi, I'm Ole Reissmann, a journalist who builds things. I'm the first Director of AI at SPIEGEL. Before that: podcasts, news product development, platform strategy. I write about AI and journalism and send a newsletter you might enjoy.
A six-step toolkit for newsrooms that want to reach younger audiences: localize research, prototype an idea, set AI guardrails, write a road map, stress-test the MVP, then ship and read the signals. Built from an eight-month sprint with 12 US newsrooms. (News Product Alliance)
The SPUR coalition now has 80 members, after The New York Times, Funke, The Economist and 40 others joined. The goal: common rules for AI content deals, before the AI companies just write those rules themselves. (Charlotte Tobitt, Press Gazette)
Tarbell Center for AI Journalism, a nonprofit close to the AI safety movement, will spend $10 million on AI coverage by the end of 2027. Kevin Roose and Casey Newton get $480k for their new NPR video podcast “Machine Gods.” (Max Tani, Semafor)
Search often runs in two steps: grab candidates, then let a model reorder the top results. Doug Turnbull tried a new idea for the second step: an LLM turns each query into about ten yes/no questions, and a small decision model like Jev checks every result against them.
LLM Wiki v2 is another ready-to-use take on Karpathy’s LLM Wiki idea: drop in documents, and it files them into folders of Markdown notes, then answers questions with citations. It’s similar to SecondBrain, but a small classifier (hello, Jev) makes the filing decisions, so the big model isn’t called for every step.
PhotoCraft is an open-source Photoshop rebuild: layers, masks, adjustment layers, type, vectors, real PSD files, and an MCP server so AI agents can drive it too. Early alpha.
Adding “think step by step” in prompts can yields better outputs. Anthropic says that giving the chatbot a role like “data scientist” makes for different results, because a data scientist might see different things in data. Now, models are not using language, but a numerical representation: tokens. 17509, 656, 5983 is “step by step” tokenized...
The JournalismAI Festival is back in November, and registration is open. London in person or free online. Sessions range from AI agent workflows for investigations to what media IP is actually worth.
I was invited to speak at the Mediatech Hub Conference in Babelsberg, Potsdam. Here’s my script, more or less what I said on stage, and some links. Hi, thanks you for having me. Let’s start with a shoutout: Next week, "Droneland" starts streaming on Telekom’s MagentaTV. It's set in a near future where artificial intelligence...
Is Claude getting dumber after launch? livenerf runs and ranks the same task over and over again to find out.
FLUX 3 Image lets you compose from scratch with bounding boxes, make multi-element edits, render in 4K, and let agents plan layouts.
Audionaut is a free, open-source multitrack audio editor that Claude and other AI agents can control via MCP.
Contextual Retrieval cuts RAG failure rates by prepending chunk-specific context before embedding, so your knowledge base actually knows what it’s talking about. (Anthropic)
Funny thing: Google serves AI-generated answers through AI Overviews and AI Mode, but tells website owners to “manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing.”
If you want the internet to be a wondrous place for quiet, odd and poetic things, you might just have to write some HTML:
