Over the past week, the AI hardware news I've been tracking adds up to more than $610 billion in capital deployed globally — in just seven days. Not valuations. Not market cap. Actual capital expenditure commitments. Korea $550B, Japan $6B, Qualcomm's new accelerator, Kawasaki Heavy Industries' $1B AI infrastructure bond — this round of moves has already surpassed the wildest half-year of the 2000 dot-com bubble in scale. But this time the money isn't flowing into web pages. It's flowing into chips, memory, and power. Watching all of this over the past few days, I've been thinking: for investors and for builders like us making products on top of AI, what does this gamble actually mean? The Real Story Behind AI Training Bottlenecks: From GPU Scarcity → Memory Scarcity → Power Scarcity Honestly, everyone watches AI through the lens of models, but the real bottleneck was never the models — it's been the hardware. From 2023 to 2025, the bottleneck shifted from GPU scarcity to memory scarcity, and is now pushing toward power scarcity. When GPUs were tight, everyone scrambled for H100s and NVIDIA raked it in — but the part that actually throttled the H100 wasn't the GPU core, it was the HBM high-bandwidth memory. On the B200, the HBM3E stacked on top has its capacity locked up entirely by NVIDIA at SK Hynix, while Samsung is chasing hard but its yields can't keep up. That's why South Korea just committed $518B to build 4 memory fabs plus $52B for the central regions, totaling $550B ( TechCrunch ). This isn't just about filling upstream capacity — the key is that Samsung + SK Hynix are trying to flip themselves from being NVIDIA's downstream suppliers into becoming the dominant players in AI hardware. Why did downstream hardware investment kick off so late? Because for the past two years people were still watching and waiting to see if "this AI hype cycle would cool down again." By 2026, GPT-6, Claude 4, and Gemini 3 are all live, inference costs have come down, user numbe

If you've ever wanted an AI that doesn't just chat but actually does things — queries databases, calls APIs, makes decisions, and learns from results — you're in the right place. In this tutorial, I'll show you how to build production-ready AI agents using Solon 4.0's ReActAgent . By the end, you'll have built an agent that can reason through complex problems, use external tools, and adapt its behavior based on real-world feedback. What Makes ReActAgent Different? Traditional LLMs are great at generating text, but they hit a wall when they need to interact with the real world — checking a database, fetching live data, or performing calculations. ReActAgent (Reason + Act) breaks through that wall. It implements a cognitive loop: Thought → Action → Observation → (repeat or finish) The agent thinks about what to do next, acts by calling a tool, observes the result, and decides whether to continue or deliver the final answer. This isn't just theory. Solon's ReActAgent has been used in production for automated customer support, intelligent data analysis, and multi-step workflow automation. 1. Adding the Dependency First, add the solon-ai-agent module to your project: <dependency> <groupId> org.noear </groupId> <artifactId> solon-ai-agent </artifactId> </dependency> Note : If you're using Solon's parent POM, the version is managed automatically. Otherwise, use the latest Solon version. 2. Building a ChatModel (The Agent's Brain) Every agent needs a "brain" — a ChatModel that powers reasoning. Let's build one using the fluent API: import org.noear.solon.ai.chat.ChatModel ; ChatModel chatModel = ChatModel . of ( "https://api.moark.com/v1/chat/completions" ) . apiKey ( "your-api-key-here" ) . model ( "Qwen3-32B" ) . build (); You can also configure it via YAML and inject it: solon.ai.chat : demo : apiUrl : " http://127.0.0.1:11434/api/chat" provider : " ollama" model : " llama3.2" @Inject ( "${solon.ai.chat.demo}" ) ChatConfig chatConfig ; ChatModel chatModel = ChatModel . o

📖 Article original : GitHub Gist Un guide technique par Mohamed ben mallessa Le problème Recevoir un dossier de 500 fichiers HEIC à convertir en WebP pour un site web est une situation courante pour tout développeur. Les solutions traditionnelles ont leurs limites : ImageMagick nécessite des codecs spécifiques, les convertisseurs en ligne sont limités en taille, et le traitement manuel est exclu à cette échelle. La solution Photopea (Photoshop gratuit dans le navigateur) supporte nativement tous les formats d'image courants. En l'utilisant comme moteur de conversion piloté par script, on obtient un pipeline batch rapide et fiable. Formats supportés Entrée Sorties possibles HEIC / HEIF JPG, PNG, WebP JPEG WebP, PNG, PSD PNG JPG, WebP WebP PNG, JPG PSD PNG, JPG, WebP SVG PNG, JPG TIFF PNG, JPG, WebP Pipeline Dossier source (500 HEIC) → Photopea → Dossier sortie (500 WebP) Le script préserve la structure des sous-dossiers, applique le redimensionnement et la qualité configurés, et livre les fichiers organisés. Paramètres typiques --format webp # Format de sortie --quality 80 # Qualité (1-100) --resize 1920 # Redimensionnement (côté long) --output ./web/ # Dossier de destination Avantages Un seul outil pour tous les formats d'entrée Aucun codec à installer (Photopea gère tout nativement) Gratuit et sans abonnement Local — les fichiers ne quittent pas votre machine Structure préservée — l'arborescence est conservée Mohamed ben mallessa — Full-stack developer & solutions B2B 🔗 GitHub · LinkedIn opensource #webp #python #tutorial 💻 Vous avez un projet technique ? Développement full-stack, automatisation IA, solutions B2B sur mesure. 🔗 GitHub 💼 LinkedIn 🎨 Behance Article initialement publié sur GitHub Gist

A gente sempre ouve falar que o sistema operacional impede que um processo veja a memória do outro ou que o programa fale diretamente com o hardware, mas normalmente não explicam o "como". Eu sempre achei isso meio mágico até que eu resolvi ir atrás da resposta, e é bem interessante. Vou me basear na arquitetura x86, mas é provável que outras arquiteturas sejam parecidas. O problema: a CPU Pra CPU não existe processo, kernel, sistema operacional. Existe só endereços de memória de onde ela lê a próxima instrução e executa. Se a CPU pode falar direto com a RAM, SSD, teclado, mouse, tela... O que me impede de escrever um programa pra ler suas senhas e tokens direto da RAM? Ou de ler arquivos e alterar arquivos sensíveis direto no SSD? Por outro lado, se o kernel fiscalizasse cada instrução que da CPU antes dela executar, isso seria extremamente lento... Outro problema: os interrupts Se a CPU só executasse sequencialmente, seu sistema poderia executar várias coisas e esquecer de checar se uma tecla foi apertada, se o mouse mexeu, etc... Então certos eventos interrompem o que quer que a CPU esteja fazendo para serem tratados assim que possível. Alguns exemplos de interrupt são: Teclas do teclado pressionadas ou soltas Botões e movimento do mouse Timers Operações de disco assíncronas Pacotes de rede recebidos/transmitidos Uma solução: rings Os processadores da arquitetura x86 tem o esquema de rings. Pense em rings como grau de limitação. Ring 0 significa limitação zero, ou seja, acesso a todas as instruções da CPU e consequentemente acesso total ao hardware e memória. O kernel roda em ring 0, ou kernel mode. O kernel assim que é carregado configura todos os interrupts handlers da CPU para executar o handler apropriado do kernel, em kernel mode, claro. Em ring 3 a CPU fica limitada e não pode fazer instruções consideradas privilegiadas. E obviamente em ring 3 a CPU não consegue se colocar em ring 0 sozinha, pois dessa forma qualquer programa conseguiria se pôr em ring 0. O

The Problem That Wouldn't Leave Me Alone Pakistan has 220 million people. A functioning legal system. Hundreds of Acts, ordinances, and constitutional provisions that technically protect every citizen. Almost nobody can use them. The median lawyer's consultation fee in Karachi is more than what many families earn in a week. Legal aid is understaffed and geographically concentrated in major cities. And the laws themselves? Written in English — a language most of the population reads functionally at best, and doesn't speak at home at all. So when a landlord illegally locks someone out. When a factory worker gets fired without severance. When a woman wants to know her inheritance rights. When a tenant needs to understand what "Section 16 of the Rent Restriction Ordinance" actually means for their specific situation — they either find a lawyer they can't afford, ask someone who doesn't really know, or quietly give up. This isn't a knowledge problem. It's an access problem. I'm a CS student at Sukkur IBA University in interior Sindh — not Karachi, not Islamabad. The kind of city where you feel the gap between what the law says and what people actually know it says every single day. That gap is where HAQ started. HAQ is an Arabic and Urdu word. It means right — as in, what is rightfully yours. The name felt important. The Core Idea: Ask the Law, Get the Law There's a specific failure mode with AI and legal questions that drove every design decision I made, and it's worth naming clearly. Standard LLMs — any of them — will answer legal questions confidently. They'll cite "Section 144" or "the Transfer of Property Act" with total authority. They are often wrong. Sometimes subtly: the section exists but doesn't say what the model claims. Sometimes obviously: the Act doesn't apply in that province. Always uncitable: the user has no way to verify without finding the source themselves. For an accessibility tool, a confidently wrong answer isn't neutral. It's actively dangerous.

If you've ever tried integrating a large language model (LLM) into a Java application, you've probably written a lot of boilerplate: HTTP clients, JSON parsing, streaming handling, session management. Solon 4.0's ChatModel abstracts all of that away with a clean, builder-oriented API. In this guide, I'll walk through building real, working AI features using ChatModel — from a simple chat call to a streaming chatbot with conversation memory. 1. What Is ChatModel? ChatModel (package org.noear.solon.ai.chat ) is a unified LLM client in Solon's AI ecosystem. Instead of writing raw HTTP calls for different model providers, you use a single API that supports: Synchronous calls — one-shot request, full response Streaming calls — reactive streaming via Project Reactor ( Flux<ChatResponse> ) Tool/Function Calling — let the LLM invoke your Java methods Chat Sessions — automatic conversation memory Multi-modal messages — text, images, audio Dialect adaptation — works with OpenAI, Ollama, Anthropic, Gemini, DashScope, and more The best part? It uses a dialect pattern — you point it at any compatible LLM endpoint, and it adapts automatically. 2. Setting Up Add the dependency to your pom.xml (no parent POM needed — Solon works standalone): <dependency> <groupId> org.noear </groupId> <artifactId> solon-ai </artifactId> <version> ${solon.version} </version> </dependency> This pulls in all built-in dialects (OpenAI, Ollama, Gemini, Anthropic, DashScope). 3. Configuration 3.1 Via YAML (Recommended) solon.ai.chat : demo : apiUrl : " http://127.0.0.1:11434/api/chat" # Full URL, not baseUrl provider : " ollama" # Dialect identifier model : " llama3.2" # Model name headers : x-demo : " demo1" Then create a @Bean to get a ready-to-use ChatModel : import org.noear.solon.ai.chat.ChatConfig ; import org.noear.solon.ai.chat.ChatModel ; import org.noear.solon.annotation.Bean ; import org.noear.solon.annotation.Configuration ; import org.noear.solon.annotation.Inject ; @Configuration public cla

Hey dev community! I don't have a PC yet, so I challenged myself to build an open source ethical hacking & "security audit" tool called ImCurvin' completely on my phone screen. Here is my honest background: I just started learning programming 6 months ago (yes, January 2026!). Since then, I've been fully focused on coding. Just 2 weeks ago, I decided to jump into cybersecurity. To be honest, I hate memorizing dry theories. I prefer focusing purely on the logic and finding cracks in systems. While I am a bit lazy with documentation, I am doing my best to learn all the official "technical terms" along the way! The project currently stands around 1,155 lines of code (993 Bash, 162 Python), i want it to be minimalist. The tool is fully modularized with separate payload directories, though the overall formatting might look a bit messy since every single character was typed on a smartphone touch screen. I want to keep this post super short, so please check out the full features, code documentation, and contribution guidelines directly inside my repository: Check out https://github.com/Skokoo/ImCurvin Feel free to check it out, and please submit a Pull Request or open an issue if you want to help a beginner clean up and optimize the code!

Half a year ago, I wanted to see for myself what can we currently have with local LLMs. I went down the rabbit hole, learned quite a lot in the process, and shared my results in an article . The results were pretty discouraging: even with 32 GB VRAM, the best models I could run were both too slow and too dumb. At the same time, what you could get for free from inference providers was actually decent - and much faster. I remember my conclusion: "Let's wait for the next generation of models, which looks very promising. If we can run something comparable to full-size Qwen3-Coder-480B locally, that would be year of the Linux Desktop age of fully capable local LLMs. And now this day has arrived. Models Half a year later, I'm revisiting this question. And this time, the whole situation has turned upside-down. Almost none of the providers still have free tier, and anything that's still free is barely good enough even for the simplest tasks. And is rate-limited all over. And on the local side, the next Qwen lineup is out. So, that's what I'm going to be looking at. Once again, I have two RX6800's, 16 GB each, and 64 GB RAM. On one hand, this is more VRAM than any "normal person" can have with one GPU - unless you've got something specifically for AI, like an unified-memory Mac or a DGX Spark. On the other hand, RX6800 is "pre-AI" - anything newer will have much better performance thanks to tensor processors. Qwen3.6-27B : This is a dense model, so basically you can't run it at all on anything less than 32 GB VRAM. It's the slowest one, but also the best one if you can run it. Its accuracy is claimed to be on par with Claude 4.5 Opus, and better than Qwen3.5-397B-A17B . This is what I've been waiting for. It runs reasonably fast on my setup, so it's very much usable both in terms of performance and accuracy. Qwen3.6-35B-A3B : This one is MoE, and it's pretty small, so it's the fastest one. It's good for anything that doesn't require too much (i.e. for agentic tasks that don'

GitHub said this week that Copilot CLI, when it runs inside a GitHub Actions workflow, will accept the built-in GITHUB_TOKEN for authentication. Per the July 2 changelog, the previous path required creating and storing a personal access token. The operational read is small and precise: one fewer human-owned credential to mint, rotate and inherit. The exact scope of the change The changelog covers a narrow surface. It applies to Copilot CLI when invoked from a GitHub Actions workflow, and it swaps the required credential from a PAT to the workflow's ambient GITHUB_TOKEN . GitHub does not describe changes to how Copilot CLI authenticates outside Actions, and this post will not extrapolate to those contexts. If your Copilot CLI usage lives on a developer laptop or in another CI system, nothing in this announcement moves for you. Why the PAT was the wrong credential to leave in the loop A personal access token has almost none of the properties you would want from an automation credential. It does not expire on a job boundary. It carries a person's identity, not the workflow's. It sits in Actions secrets long enough to outlive the engineer who created it. And its scopes were chosen by that engineer, at that moment, often wider than the job actually needs. GITHUB_TOKEN is the opposite shape. Actions mints it at the start of a job, scopes it through the workflow's permissions: block, and revokes it when the job ends. If the token leaks, the window for abuse is the runtime of the job, not the years until somebody remembers to rotate it. When the person who wrote the workflow leaves, the pipeline does not silently break because a token expired with their account. For scripted Copilot CLI calls that had to be wrapped in a PAT, that is the whole win. The tool authenticates against the workflow instead of against a human. Wiring it up The workflow-side pattern is the same one every GITHUB_TOKEN -consuming step already follows: declare permissions: explicitly at the job level, k

We’ve all been there. You start your morning feeling like a Productivity God, sitting straight and typing at 120 WPM. Fast forward four hours, and you've morphed into a literal shrimp, face inches away from the monitor, hunting for a missing semicolon. 🦐 In this era of remote work, real-time posture correction and computer vision for health have become more than just "cool projects"—they are spinal lifesavers. Today, we’re going to build a desktop application using MediaPipe , WebRTC , and Electron that monitors your neck angle and sends a desktop notification the moment you start slouching. By leveraging MediaPipe Pose and TensorFlow.js , we can calculate the Forward Head Posture (FHP) ratio with surgical precision directly in the browser environment. The Architecture 🏗️ Before we dive into the code, let’s look at how the data flows from your webcam to that "Sit up straight!" notification. graph TD A[Webcam Feed] -->|MediaStream| B(WebRTC API) B -->|Video Frames| C[MediaPipe Pose Model] C -->|Landmarks| D{Geometry Engine} D -->|Calculate Ear-Shoulder Angle| E{Threshold Check} E -->|Angle > 30°| F[Electron Main Process] F -->|Trigger| G[System Desktop Notification] E -->|Healthy| H[Continue Monitoring] style G fill:#f96,stroke:#333,stroke-width:2px Prerequisites 🛠️ To follow along, you'll need the following tech stack: MediaPipe Pose : For high-fidelity body tracking. WebRTC : To capture the video stream from your webcam. Electron : To wrap our logic into a desktop app that runs in the background. TensorFlow.js : The backbone for running ML models in JavaScript. Step 1: Setting up the Video Stream (WebRTC) First, we need to grab the camera feed. In a modern browser environment (or Electron's Chromium), we use navigator.mediaDevices.getUserMedia . async function setupCamera () { const videoElement = document . getElementById ( ' input_video ' ); const stream = await navigator . mediaDevices . getUserMedia ({ video : { width : 640 , height : 480 }, audio : false }); v

If you've ever built a feature that extracts text from PDFs, an Arabic-speaking user has probably filed this bug: "the words come out in reverse order." Not the letters — the words . Every line reads last-word-first. I spent the better part of a year fixing this class of bugs while building Confileo , a free PDF toolkit with first-class Arabic support. Here's what's actually going on, because almost every explanation online is wrong or incomplete. The four distinct failure modes People say "Arabic breaks" as if it's one bug. It's four: 1. Visual vs logical order (the reversed-words bug) A PDF doesn't store text the way a Word file does — it stores positioned glyph runs : "paint these shapes at these coordinates." For left-to-right scripts, the paint order happens to match the reading order, so naive extraction works by accident. Arabic is right-to-left. Many PDF generators emit the glyph runs in visual order — the order they appear on screen, left to right. A naive extractor concatenates the runs as stored and produces every line word-reversed. The text was never "reversed" in the file; your extractor just assumed paint order == reading order. Fix: reconstruct logical order using glyph positions + the Unicode Bidirectional Algorithm (UAX #9), not the content-stream order. Libraries like PyMuPDF already return text in logical order — a common mistake is "fixing" that output by reversing it again, which is how you get double-reversed text. Rule of thumb: never reverse Arabic yourself. If it looks backwards, your rendering layer lacks bidi support; the data is usually fine. 2. Disconnected letters (the ransom-note bug) Arabic letters are contextual: ع renders differently in initial, medial, final and isolated positions, and letters join. That joining is applied at render time by a shaping engine (HarfBuzz being the standard). If any step of your pipeline round-trips text through a non-shaping renderer — a canvas library, a barebones PDF writer, an image caption filter

Second in a series on using AI to review authorization — not to spray reports. Companion reference: AuthZ Smell Catalog . 1. Why AuthZ review is not vulnerability spraying The cheapest thing an AI can do in security is generate suspicion. Point a model at a codebase and it will hand you fifty "possible IDORs" before you finish your coffee. Almost all of them are wrong — guarded three lines up, scoped at the data layer, or protected at a boundary the model never saw. That flood is exactly why several bug bounty programs spent 2026 tightening or pausing: they were drowning in confident, plausible, wrong reports. So this review inverts the usual loop. The AI's job is not to find bugs — it is to over-generate hypotheses cheaply . My job is to kill them. What survives that killing is the only thing worth a human's time, and the record of what died is more useful than the record of what lived. The artifact of an honest review is therefore not a finding. It's a kill table . 2. Target and scope Target: Ory Kratos — an open-source identity and user-management server (login, registration, recovery, verification, sessions, self-service settings). Source-available, Apache-2.0. Why Kratos: it is exactly the shape where authorization goes wrong — multiple identities, a public API and an admin API, and (in Ory's hosted product) multi-tenancy. If a boundary is fragile, this is where it shows. Scope of this write-up: source reading only , on the public repository, single-tenant OSS build. No hosted target was touched. Nothing here is an undisclosed finding — the point is the method and the boundary design , and where relevant, how the design held against the hypotheses I tested. This maps to the reproduction tiers we track: everything below is repo_only , and I say so explicitly rather than implying it reaches a live product. What this review does and does not claim. In this limited, repo-only review, the hypotheses I tested were killed. This is not a claim that Kratos has no vulner