AI News Today

<h2> Your repo is already your agents' context, whether you designed it on purpose or not </h2> <p>That sentence took me a while to understand. In this post I'll save you the trip.</p> <p>It was October 2025, working in Skyward's monorepo with AI agents every day. And every day the same routine: I'd tell the agent in the prompt "don't use this", "don't do it this way", "reuse the component that already exists". I wrote it down. I repeated it.</p> <p>The agent did exactly what I told it not to do

<h2> Tu repo ya es el contexto de tus agentes, lo hayas diseñado a propósito o no </h2> <p>Esa frase me costó entenderla. En este post te ahorro el camino.</p> <p>Era Octubre del 2025, trabajando en el monorepo de Skyward con agentes de IA todos los días. Y todos los días la misma rutina: le decía en el prompt al agente "no uses esto", "no hagas esto así", "reutiliza el componente que ya existe". Lo escribía. Lo repetía.</p> <p>El agente hacía exactamente lo que le dije que no hiciera.</p> <p>No

Article URL: https://finance.yahoo.com/technology/ai/articles/ai-holdouts-tech-face-3-095310222.html Comments URL: https://news.ycombinator.com/item?id=48593519 Points: 3 # Comments: 1

<h2> Architecture Overview </h2> <p>Clioloop's Agentic Fusion is not just "run the same prompt 5 times and pick the best." It's a structured pipeline where different models play different roles, with strict security boundaries between them.</p> <h2> The Pipeline </h2> <h3> Step 1: Planning Phase </h3> <p>When you run <code>/fusion</code>, up to 5 <strong>planner models</strong> are dispatched in parallel. Each planner:</p> <ul> <li>Receives your original prompt and context</li> <li>Has <strong>r

Article URL: https://www.bloomberg.com/news/articles/2026-06-18/ai-costs-what-shift-from-flat-rate-to-token-pricing-model-means-for-industry Comments URL: https://news.ycombinator.com/item?id=48593411 Points: 2 # Comments: 0

<p>Build-in-public. This is the real architecture behind running ~50 local AI agents on 6GB of VRAM — one GPU lock, an eviction watchdog, a resource governor, and a model router. Originally posted on my blog.</p> <p>The question I get most often is some version of "there's no way you run that many agents on a 6GB laptop GPU." The honest answer: not the way you're picturing it. I don't run 50 models at once. I run one model at a time, very deliberately — and most of the engineering is about sched

<p>On June 12, 2026, enterprise developers using the Codex API started seeing an unfamiliar response header: <strong><code>X-Model-Version: kindle-alpha</code></strong>. It appeared on a subset of requests for roughly 18 hours, then vanished. That's the release candidate for GPT-5.6 — OpenAI's next flagship model — leaking through the staging layer. OpenAI's Chief Scientist publicly called the upcoming release "a meaningful leap" the following day. By OpenAI's historically understated communicat

<h2> The Problem </h2> <p>Most AI assistants give you one model's answer. If it's wrong, you catch it or you don't. If you use a cheap model, quality drops. If you use a frontier model, you pay frontier prices for everything — even a simple file rename.</p> <h2> What is Agentic Fusion? </h2> <p>When you run <code>/fusion</code>, a panel of models collaborates on your task:</p> <ol> <li><p><strong>Planners</strong> (up to 5): Read-only models that research and propose routes in parallel. They fig

<pre style='white-space:pre-wrap;width:81ex'>Merge branch 'main' of https://github.com/openclaw/openclaw * 'main' of https://github.com/openclaw/openclaw: (82 commits) fix(e2e): validate chat tools body limit fix(e2e): honor chat tools body limit fix(e2e): validate chat tools timeout fix(e2e): give cleanup smoke build heap headroom fix(e2e): validate plugin lifecycle limits refactor(auto-reply): add lifecycle storage seams (#93685) fix: preserve pending subagent completion announces (#94349) fix

Hi all! I was recently doing a live stream session at Build and created the start of a tool called Flashback. I wanted a way for agents to look through the history of popular design so you could reference different trends throughout time. I didn’t finish on the stream, but I kept coming back to the idea so ended up shipping it as a GitHub pages powered Skill and human browsable web site. I think it’s really interesting to see how design has evolved since 1900, and I learned a lot in this process

Article URL: https://github.com/sricola/drydock Comments URL: https://news.ycombinator.com/item?id=48593356 Points: 1 # Comments: 0

<h2 id="abstract">Abstract</h2> <p>Artificial intelligence (AI) agents are moving from experiments into everyday engineering workflows. They can read code, call application programming interfaces (APIs), run tests, create merge requests, answer Slack messages, and keep long-running state. That makes them useful, but it also changes the risk model - especially as agents get more autonomous in their use of tools. An agent with network access, credentials, tools, and memory is no longer just a chat

This is 100 Years From Now. Once a week we skip a century and try to picture what life actually looks like when the stuff we're building now has had time to settle in. This week: the war that won't have a country in it.

Article URL: https://artificialanalysis.ai/articles/aa-briefcase Comments URL: https://news.ycombinator.com/item?id=48593225 Points: 6 # Comments: 1

Scientists are exploring new algorithms, hardware, and computing methods to lower AI’s power demands. Strategic siting of data centers and other steps to increase green energy use are also key. The post How to Tame AI’s Voracious Appetite for Energy appeared first on SingularityHub .

Amazon SageMaker AI provides fully managed real-time inference hosting for machine learning models. You deploy a model to a SageMaker endpoint backed by one or more compute instances, and SageMaker handles provisioning and scaling. SageMaker supports multiple endpoint architectures. This post focuses on the two most relevant to generative AI workloads with detailed observability: Single-model endpoints (SME) and Inference component (IC) endpoints.

337+ MCP tools — weather, crypto, finance, web search, DNS, colors, AI chat, text analysis, calculators, and more. $5/500 calls, each call 1 credit. Comments URL: https://news.ycombinator.com/item?id=48593013 Points: 1 # Comments: 0

As software engineers, we've definitely seen the best use case for LLMs so far. Coding agents have shown us what these models can do in an area we can actually verify. But lately I've found myself comparing my doctor's notes against LLM answers more and more. Curious how others are using LLMs, and relying on them, for non-engineering stuff. Comments URL: https://news.ycombinator.com/item?id=48593007 Points: 2 # Comments: 1

Anthropic announced a potentially game-changing new feature for users of Claude Code on the Claude Team and Enterprise subscription plans: Artifacts . This update turns a Claude Code session's work into a live, interactive, and shareable, custom HTML webpage, allowing a Claude Code user to plug in live code, multiple data sources, and have it surface on an interactive URL that they can send to other teammates — be it a dashboard, an app design, or some other product meant for internal usage. The

Article URL: https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world Comments URL: https://news.ycombinator.com/item?id=48592899 Points: 1 # Comments: 0

Hello, I've created this project to simplify the provisioning of Hermes on Hetzner VPS. The goal is adopting good security defaults to deploy Hermes Agent on Ubuntu 24 LTS + Tailscale. I've added also backups and the optional installation of Mnemosyne as light memory provider. I'm currently working on it, any feedback is appreciated Comments URL: https://news.ycombinator.com/item?id=48592879 Points: 1 # Comments: 0

Article URL: https://api.intelica.dev Comments URL: https://news.ycombinator.com/item?id=48592817 Points: 2 # Comments: 0

With AI, it's faster than ever to ship a marketing site... but most of what gets generated is slop that was never built to be found. Plus the tools meant to catch that fall short: most SEO auditors cost money, don't play nicely with your agents, or tell you what's wrong without telling you how to actually rank for SEO and GEO (Generative Engine Optimization: being cited by AI search like ChatGPT, Perplexity, and Google AI Overviews). crawlie fixes that! It's 100% free, it's local-first, it's age

Article URL: https://www.massgeneralbrigham.org/en/about/newsroom/press-releases/evaluating-ai-performance-for-everyday-patient-care Comments URL: https://news.ycombinator.com/item?id=48592716 Points: 1 # Comments: 0

Article URL: https://www.news.iastate.edu/news/writing-ai-demands-more-thought-students-not-less Comments URL: https://news.ycombinator.com/item?id=48592621 Points: 1 # Comments: 1

Article URL: https://gizmodo.com/a-federal-regulator-wants-to-fast-track-ai-data-centers-onto-the-power-grid-2000773803 Comments URL: https://news.ycombinator.com/item?id=48592535 Points: 1 # Comments: 0

I'm the CTO of a small NGO (10 people, 3 devs). Each person has an account with one of the leading AI companies. I'm looking for an easy, reliable and cheap way to allow them to connect to external MCP servers without having the hassle of setting everything up. What I want is a simple MCP server that I can show them how to connect, which aggregates access to multiple MCP servers I configure (Slack, Google Drive, Asana, and others). Kind of an "MCP Gateway". I tried: * Pipedream.com: Great to agg

Article URL: https://www.palmier.io/ Comments URL: https://news.ycombinator.com/item?id=48592416 Points: 1 # Comments: 0

Article URL: https://www.callercallsback.com/p/ceos-heres-how-to-not-screw-up-your Comments URL: https://news.ycombinator.com/item?id=48592239 Points: 1 # Comments: 0

Article URL: https://subq.ai/ Comments URL: https://news.ycombinator.com/item?id=48592153 Points: 2 # Comments: 0

Article URL: https://knowablemagazine.org/content/article/technology/2026/what-shapes-ai-moral-decisions Comments URL: https://news.ycombinator.com/item?id=48592100 Points: 1 # Comments: 0

Article URL: https://sycloop.com Comments URL: https://news.ycombinator.com/item?id=48592065 Points: 1 # Comments: 0

Article URL: https://radiologyai.com/ Comments URL: https://news.ycombinator.com/item?id=48592044 Points: 2 # Comments: 0

<!-- SC_OFF --><div class="md"><p>I maintain cuTile Rust and just posted the paper "Fearless Concurrency on the GPU." </p> <p>As more GPU code gets AI-generated, the bottleneck moves from writing it to trusting it. cuTile Rust lets you write or generate GPU kernels whose memory safety and data-race freedom are verified by the compiler, through Rust's ownership and borrow checking. You get those guarantees by construction. It's a tile-based programming model that lowers to CUDA Tile IR, carrying

Announcing the general availability of Amazon Elastic Compute Cloud (Amazon EC2) G7 instances, delivering high performance GPU acceleration for AI inference, graphics, and data analytics workloads.

Startup Baseten is reportedly close to finalizing a $1.5 billion round at a $13 billion as the “inference gold rush" marches on.

Anthropic still can’t distribute Claude Mythos or Fable 5 after running afoul of the Trump administration. But no one can say exactly what the company did wrong.

Article URL: https://www.ft.com/content/9f063b07-da39-4feb-92ab-ee0f91385c62 Comments URL: https://news.ycombinator.com/item?id=48591546 Points: 5 # Comments: 1

I've had more than six consultations with top veneer dentists in NYC, Turkey, and Mexico. All six gave me a different recommendation. Not one of them offered to show me what I'd actually look like with veneers. So I tried ChatGPT, Gemini, and Grok. They all changed my face, and none of them let me compare options to figure out which smile I'd actually want. So I built it myself. Submit a photo and get back ten veneer options in under 60 seconds. Upload as many photos as you want and see how you'

Article URL: https://github.com/anthonystaiger8-bit/memory-model Comments URL: https://news.ycombinator.com/item?id=48591389 Points: 2 # Comments: 0