PulseAugur / Brief
EN
LIVE 21:42:21

Brief

last 24h
[21/21] 221 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. SIGNIFICANT · Simon Willison English(EN) · · [2 sources]

    The memory shortage is causing a repricing of consumer electronics

    A significant shortage in memory production is driving up the cost of consumer electronics, particularly impacting the sub-$100 smartphone market. This shortage is attributed to the increasing demand for High Bandwidth Memory (HBM) used in AI data centers, which now consumes a much larger portion of wafer capacity compared to traditional DDR and LPDDR memory. Memory manufacturers are prioritizing HBM due to its profitability and the lessons learned from past industry consolidation, leading to constrained production of memory for everyday devices. AI

    IMPACT AI's demand for HBM is causing a memory shortage, leading to higher prices for consumer electronics and impacting the availability of affordable smartphones.

  2. TOOL · Simon Willison Italiano(IT) · · [3 sources]

    datasette-fixtures 0.1a0

    Simon Willison has released alpha versions of Datasette 1.0, introducing a new customizable "Jump to..." menu and a "datasette-agent" plugin. The agent plugin integrates with the new menu to provide an interface for starting AI chats. Additionally, a "datasette-fixtures" plugin has been released, which aids in creating fixture databases for plugin test suites. AI

    IMPACT Enhances data exploration tools with AI chat integration, potentially improving developer workflows.

  3. TOOL · Simon Willison English(EN) ·

    Mad House — Usborne Creepy Computer Games

    Simon Willison recreated a classic 1980s computer game, "Mad House," using Claude. The game, originally from Usborne's "Creepy Computer Games" book, was typed into a Commodore 64 by Willison as a child. He fed the book's PDF into Claude, which then generated interactive JavaScript and HTML code for a mobile-friendly, retro-styled version of the game. AI

    Mad House — Usborne Creepy Computer Games

    IMPACT Demonstrates LLMs' ability to generate functional code from scanned documents, enabling recreation of classic software and interactive experiences.

  4. RESEARCH · Simon Willison English(EN) · · [3 sources]

    FTC to Require Cox Media Group, Two Other Firms to Pay Nearly $1 Million to Settle Charges They Deceived Customers About “Active Listening” AI-Powered Marketing Service

    The FTC has ordered Cox Media Group and two other firms to pay nearly $1 million to settle charges of deceiving customers about an AI-powered marketing service. These companies falsely advertised an "Active Listening" service that they claimed used smart devices to capture real-time voice data for targeted advertising. In reality, the service involved reselling email lists at a markup and did not actually listen to conversations or use voice data. AI

    IMPACT This action highlights regulatory scrutiny on AI marketing claims, potentially impacting how companies advertise AI-powered services.

  5. TOOL · Simon Willison Italiano(IT) · · [3 sources]

    datasette-llm 0.1a8

    Simon Willison has released updates for his open-source Python tools, datasette-llm and datasette-llm-accountant. Version 0.1a8 of datasette-llm fixes a bug related to collecting chains of responses within the llm_prompt_context hook. The datasette-llm-accountant tool, now at version 0.1a4, also addresses a bug concerning the tracking of response chains. AI

    IMPACT Minor tooling improvement for developers working with LLMs.

  6. COMMENTARY · Towards AI English(EN) ·

    Karpathy Said Kill Prompt Engineering. Here’s What He Actually Meant.

    Andrej Karpathy proposed a shift from "prompt engineering" to "context engineering," arguing that the former term inaccurately represents the complex task of building production AI applications. He clarified that real-world AI systems rely on dynamically assembling a context window from various sources like system instructions, retrieved documents, and conversation history, rather than just simple user prompts. This assembly process, which Karpathy deems the true engineering challenge, is crucial for optimal model performance and is distinct from merely crafting task descriptions. AI

    Karpathy Said Kill Prompt Engineering. Here’s What He Actually Meant.

    IMPACT Reframes the core challenge in building production AI applications from prompt phrasing to managing dynamic context assembly.

  7. COMMENTARY · Simon Willison Deutsch(DE) ·

    Quoting Armin Ronacher

    Armin Ronacher criticizes the current trend of AI-generated issue reports, which often lack clarity and accuracy due to poor prompting. He advocates for issue submissions that strictly adhere to a human-observed format: detailing the command run, expected outcome, actual outcome, and exact errors. This approach aims to cut through the noise and provide actionable information for developers. AI

    IMPACT AI-generated content is creating noise and hindering effective communication in software development.

  8. COMMENTARY · Simon Willison English(EN) · · [4 sources]

    The last six months in LLMs in five minutes

    Simon Willison presented a five-minute talk at PyCon US 2026 summarizing LLM developments since November 2025. Key advancements included significant improvements in coding agents, which became reliable for daily use, and the emergence of 'Claws'—personal AI assistants like OpenClaw that drove sales of Mac Minis for local hosting. AI

    The last six months in LLMs in five minutes

    IMPACT Summarizes key LLM progress in coding agents and personal assistants, highlighting their increasing utility and market impact.

  9. COMMENTARY · Mastodon — mastodon.social English(EN) ·

    A quick catch up from Simon Willison about progress over the last six months in LLMs. TL;DR: Coding agents have gotten really good and the laptop-available open

    Open-source large language models that can run on laptops have significantly improved, now vastly exceeding prior expectations. Coding agents, in particular, have demonstrated remarkable progress in recent months. This advancement suggests a rapid evolution in the capabilities of locally deployable AI. AI

    IMPACT Highlights the surprising performance gains in open-source LLMs and coding agents, suggesting faster adoption of local AI capabilities.

  10. COMMENTARY · Simon Willison English(EN) ·

    On the <dl>

    Simon Willison's blog post highlights lesser-known features of the HTML `<dl>` element, drawing from an article by Ben Meyer. Key takeaways include that a `<dt>` can precede multiple `<dd>` elements and that these pairs can be semantically grouped within a `<div>` for styling purposes. The post also notes the official name change to "description lists" since 2008 and mentions ARIA labeling for accessibility, referencing Adrian Roselli's insights on screen reader compatibility. AI

  11. TOOL · Bluesky Jetstream — AI desk English(EN) ·

    Fun new feature in Datasette 1.0a30 - a "jump to" menu, also available by the "/" keyboard shortcut, which lets you type to jump to a database, table or canned

    Datasette 1.0a30 introduces a new "jump to" menu, accessible via a keyboard shortcut or a slash command. This feature allows users to quickly navigate between databases, tables, and predefined queries within the application. It also includes a plugin hook to enable extensions with custom content. AI

    IMPACT Enhances user experience for data exploration tools.

  12. SIGNIFICANT · Simon Willison Italiano(IT) · · [7 sources]

    datasette-agent-sprites 0.1a0

    Google's Gemini 3.5 Flash model, while fast, is significantly more expensive than its predecessors, with estimates suggesting a total parameter count between 250 billion and 300 billion. Despite its speed, users report that it can be prone to generating overly elaborate outputs and may struggle with precise structural corrections. Discussions on Hacker News indicate that while Gemini 3.5 Flash excels at one-shot coding tasks, its performance in long-term agentic tasks requiring tool use is less robust. AI

    IMPACT Sets a new benchmark for high-performance, high-cost LLMs, prompting careful consideration of ROI for AI operators.

  13. TOOL · Bluesky Jetstream — AI desk English(EN) ·

    I released the first alpha of Datasette Agent - a conversational AI assistant for Datasette that can answer questions about data in SQLite databases, and can b

    Simon Willison has released the first alpha version of Datasette Agent, a new AI assistant designed to interact with data stored in SQLite databases. This web application allows users to ask questions about their data and can be expanded with custom plugins. A live demo is available on agent.datasette.io. AI

    IMPACT Provides a new conversational interface for interacting with structured data in SQLite databases.

  14. TOOL · Simon Willison Italiano(IT) · · [6 sources]

    datasette-agent 0.1a1

    Simon Willison has released Datasette Agent, an extensible AI assistant designed to interact with data stored in Datasette. The agent allows users to ask conversational questions about their data, and with plugins, can generate charts or even images. It leverages models like Google's Gemini 3.1 Flash-Lite for its operations and is designed to be highly extensible with new plugins, including those for image generation and code execution. AI

    IMPACT Enhances data interaction for users of the Datasette platform by providing a conversational interface and charting capabilities.

  16. TOOL · Bluesky Jetstream — AI desk English(EN) ·

    New TIL: I figured out how to use my LLM CLI tool in a shebang line, which means you can write executable scripts in English, or hook up more complex scripts wi

    Simon Willison has developed a method to integrate LLM CLI tools directly into shebang lines, enabling the creation of executable scripts written in natural language. This technique allows for the execution of complex scripts by combining English commands with YAML templating. The innovation facilitates a more intuitive scripting experience by leveraging LLMs for command interpretation. AI

    IMPACT Enables more accessible scripting by allowing natural language commands to execute code, potentially lowering the barrier to entry for developers.

  17. COMMENTARY · Bluesky Jetstream — AI desk English(EN) ·

    To prepare for my #PyConUS lightning talk this afternoon I decided to track down ALL of the names that @openclaw has used since November, using a script against

    Simon Willison has documented the evolving naming conventions of the AI project OpenClaw. Over several months, the project has undergone multiple name changes, including Warelay, CLAWDIS, CLAWDBOT, Clawdbot, Moltbot, and most recently, OpenClaw. This tracking was done to prepare for a lightning talk at PyCon US. AI

    IMPACT Provides insight into the branding evolution of an AI project, potentially useful for understanding project identity and communication strategies.

  18. MEME · Bluesky Jetstream — AI desk English(EN) ·

    "old woman

    A user on Bluesky shared a prompt and its output from an AI model. The prompt, "old woman", generated the response "possibly damp faster than an old woman should be". This interaction was posted on May 12, 2026. AI

  19. TOOL · Bluesky Jetstream — AI desk English(EN) ·

    Wrote about today's GitLab restructuring / "workforce reduction" announcement, and ended up digging around in version control for both the GitLab and the 37sign

    GitLab announced a significant restructuring and workforce reduction, impacting its AI desk. This move involved examining version control systems for both GitLab and 37signals' employee handbooks to illustrate the author's perspective on the situation. AI

    IMPACT This event impacts AI teams within GitLab, but the core news is a corporate restructuring rather than a direct AI development.

  20. COMMENTARY · Bluesky Jetstream — AI desk English(EN) ·

    This is excellent. I particularly like the definition of the "Zombie Internet", which starts: "It’s people talking to bots, people talking to people, people cre

    Simon Willison highlights the concept of the "Zombie Internet," a term he finds particularly insightful. This phenomenon describes a digital landscape where human interaction is increasingly mediated by or directly involves AI. It encompasses scenarios ranging from people conversing with bots to AI agents interacting with humans, and even a divide between AI users and non-users. AI

    IMPACT Defines a new conceptual framework for understanding the evolving nature of online human-AI interaction.

  21. SIGNIFICANT · Mastodon — fosstodon.org Polski(PL) · · [81 sources]

    Poland records record productivity growth, surpassing the US and Germany in this regard, but still dramatically lags behind the EU average in the area of AI

    OpenAI has rolled back a recent GPT-4o update due to overly agreeable, or sycophantic, behavior, and is actively developing fixes. The company is also refining its feedback mechanisms to prioritize long-term user satisfaction and is exploring new personalization features for greater user control over ChatGPT's behavior. Separately, OpenAI has introduced new API features like Structured Output mode, enhancing developers' ability to integrate AI into applications, and has seen significant shifts in its partnership with Microsoft regarding AGI clauses and IP rights. AI

    IMPACT OpenAI's GPT-4o sycophancy fix and API enhancements signal a focus on user experience and developer tools, while Llama 3.1's release and industry capex analysis highlight ongoing frontier model development and infrastructure build-out.