Alex Gardner

OpenAI's own engineers are merging 70% more pull requests every week. This is what changed. Codex just got a massive update and it's not about chatting with AI anymore. It's about running a team of AI workers across your codebase. Here's what just dropped: → Desktop app — manage multiple agents working on multiple tasks in parallel, each in their own thread with their own context → Background automations — set a task, define a schedule, walk away — results land in a review queue while you sleep → Automated code review — Codex reviews every pull request before a human even opens it, catching hundreds of bugs per day → Skills — package your team's exact standards so every automation follows your rules automatically Here's what OpenAI's own team runs on their codebase right now: Daily issue triage overnight instead of a developer spending an hour every morning CI failure summaries so you wake up knowing exactly what broke and why Automated bug sweeps running on a schedule before humans know issues exist Codex now reviews the vast majority of their PRs before production. At Cisco, engineers saw review times drop by up to 50%. Duolingo, Vanta, Instacart — all running this in production. Not experiments. Real workflows. The question isn't whether to use this. It's how fast you get comfortable with it before it becomes the standard expectation. Here's exactly how to set it up today ↓

Yet another claude code agent for a @kaggle competition. Using Ai agents on kaggle competitions is trending now.

GLM-OCR, 0.9B model that beats Gemini on OCR benchmarks.💀 It's a 0.9B param vision-language model. supports 8K resolution, 8+ languages, and has built-in text, LaTeX, and table recognition modes. demo's - github.com/zai-org/GLM-OCR

Gonna try it myself in a moment. Everything here is spot on: - There are major problems with today’s coding agents. - Solving them takes an insane amounts of time & effort. - You must keep maintaining your solution all the time otherwise it breaks. - No quota left for code..

Comments in this post made me think of writing about Alan Turing x.com/OjasSharma276/…

The Life of Alan Turing: Alan Turing was one of the smartest people to ever exist, and honestly, modern computing owes him a massive debt He wasn’t just a computer scientist. He was: >A mathematician >A logician (someone who studies reasoning) >A cryptanalyst (codebreaker) >A philosopher >A Theoretical biologist Turing introduced the idea of the Turing Machine, a simple concept that explains how computers work. He also worked on designing one of the first computer models called the Automatic Computing Engine (ACE). During World War II, Turing worked at Bletchley Park. His job? Break encrypted German messages. Germany used a machine called Enigma, which was considered unbreakable. Turing helped design systems that cracked these codes. Later in life, Turing even explored biology. He studied how patterns in nature form, like: >Zebra stripes >Leopard spots Despite everything he did, Turing was treated horribly by his own country. In 1952, he was prosecuted for being homosexual (which was illegal in the UK at the time). Instead of prison, he chose hormone treatment (chemical castration). This had severe physical and mental effects on him. Turing died on 7 June 1954, at just 41 years old, from cyanide poisoning. It’s widely believed to have been suicide.

🐍📰 Build a Python MCP Client to Test Servers From Your Terminal Create a Python project for an MCP client that discovers server capabilities and integrates with an AI chat. #python realpython.com/python-mcp-cli…

by the time you open a slack message, runbear has already read it, pulled context from your stack, and drafted the reply. most of the work happens before you're in the thread.

CIOs are suddenly seeing a notable uptick in autonomous AI, spurred by the success of OpenClaw and enterprise-grade equivalents like Claude Cowork cio.com/article/414665… My take: It’ll all come down to how good the guardrails and oversight are.

The real shift in AI isn’t better models. It’s ownership. OpenClaw + Ollama gives you: → Full control → Zero cloud dependency → Massive context (up to 1M tokens) → Real automation with tool calling This isn’t a tool. It’s your own AI infrastructure.

ops requests don't get simpler as you scale. runbear answers, routes, and acts on them before they become bottlenecks. gets smarter every conversation because every workflow is a data point.

In case you missed it. Scientists have achieved new milestones in nuclear fusion research, bringing the technology closer to practical use. Nuclear fusion is the process that powers the Sun. It occurs when atomic nuclei combine under extremely high temperatures and pressure, releasing large amounts of energy. For decades, scientists have been trying to reproduce this process on Earth as a potential source of clean, abundant energy. To do this, researchers use devices called tokamaks donut-shaped machines that confine super hot plasma using powerful magnetic fields. Recent experiments have exceeded previous performance limits, including improvements in plasma stability, density and confinement time. These advances address some of the key scientific challenges that have long restricted fusion progress. While fusion power plants are not yet ready to generate electricity for the grid, these results show that controlled fusion is becoming more technically achievable. If future engineering challenges are solved, fusion could provide a low carbon energy source with minimal long-term waste and virtually unlimited fuel.

Yeah. But not really “MIT elites,” I wouldn’t say it that way. These guys are the pieces on the board. The people *moving them* think we are insects.

Microsoft has announced a massive restructuring of its AI leadership, officially splitting its focus between commercial product development and foundational model research. Mustafa Suleyman (CEO of Microsoft AI and DeepMind co-founder) is stepping away from the day-to-day product management of Copilot. Instead, he will focus exclusively on leading Microsoft's "superintelligence" mission to build state-of-the-art, frontier AI models in-house. Former Snap executive Jacob Andreou has been promoted to Executive Vice President of Copilot to lead this consolidated effort, reporting directly to CEO Satya Nadella.

Qwen3.5-4B is trained locally on 4GB RAM via Unsloth Studio. ✨ Optimizations. The 4B model did this by executing tool calls & web search directly during its thinking trace. Small but mighty thanks to smart distillation and training.

> be cursor > first to market for coding IDEs (with copilot) > $60 million series-a funding > made Dr Karpathy coin "vibe coding" > they are already an RL harness from day 1 > if user "Accepts Edit" - positive reward, if not user's next message is rich feedback > thats the purest form of RLRF (RL with rich feedback) > cursor tab also - pure RLVR (does user accept autocompletion? yes or no) > they shouldve been unstoppable But then... > they had one big tech-debt. they had to rely on other providers (OpenAI, Anth) coz they didnt have any competitve coding models of their own, > Sonnet and Opus costs $$$ via API > make some pricing moves that that have soured people against them > rise of competitors (claude code and now codex) > in came the terminal era: less typing, less editing > people moved from vibe coding to automated agentic coding by 2025. But then... > all this while they had enough analytics to start training their own models > open-weight coding models are already great, many of them have open licenses too for a good base > cursor models wont need to compete on general benchmarks - just basic intelligence + coding & SWE benches are all they need > did I mention they have had a banger team for a while? And today... > they have released Composer 2 now which beats Opus 4.6 and competes with GPT 5.4 high in a fraction of the cost > hopefully the usage issues will reduce because they have a good model that's optimized for their harness + runs cheap this whole thing is playing out like a movie in my head

🚨 RIP Chrome for AI agents. Someone built a headless browser from scratch that runs 11x faster and uses 9x less memory. It's called Lightpanda. Every AI agent doing web automation right now is running Chrome under the hood. That means you're spinning up a massive desktop application, stripping out the UI, and running hundreds of instances of it on a server. For something that never needs to render a single pixel. It's like renting a semi-truck to deliver a letter. Lightpanda is built differently. Not a fork of Chromium, Blink, or WebKit. Written from scratch in Zig with one goal: headless performance, nothing else. It still runs JavaScript. Still handles Ajax, XHR, Fetch, SPAs, infinite scroll, all of it. Just without dragging along 500MB of browser bloat you'll never use. And it drops straight into your existing stack: → Compatible with Playwright, Puppeteer, and chromedp via CDP → One-line Docker install → CDP server on port 9222, swap it in for Chrome in 30 seconds The use cases are obvious: AI web agents, LLM training data scraping, browser automation at scale, testing pipelines. Anything where you're paying for Chrome compute and cringing at the bill. It's still in beta and Web API coverage is growing. But at 11.8K stars it's clearly hitting a real nerve. 100% Opensource. AGPL-3.0. Link in comments.

Creativity. Powered by Gen AI. Gen AI powered by fair use. Great to see @ProgressChamber partners @Google, @OpenAI, @Midjourney all Top 5 in @CNET’s picks for the best! cnet.com/tech/services-…

I live in Eindhoven, ASML town. Heard this from folks inside... Over the years ASML promoted a lot of strong EUV people into architect roles, group leads, management tracks. Built up serious layers. Classic growth pain. Then McKinsey comes in , says cut the management layers to speed things up. So now those same high performers, real good EUV experts, who got promoted are the ones on the block. About 3400 roles targeted, mostly management. Half will be reassigned, rest gone. Big hit in Veldhoven/Eindhoven area (~1400), some in US. Unncertainty is high, unions talking, details probably land around April. At the same time, ASML is still planning massive growth. The new campus near the airport just got final green light from city council. Construction starts soon, phased build-out. Long term they talk ~20,000 new jobs in the region (first wave ~5k by 2028). And the layoff packages are subpar. Philips is also based in the same town. And they had layoffs due to serious financial issues last year. Their layoff packages were much better than what ASML is offering. What a way to kill tje company culture. Especially at a time when the company is printing money.

I am not bullish on AI-only orgs just yet. We will see increasing AI:human intelligence output ratio within orgs, but I believe human collaboration and ai orchestration should exist in the same place to make context sharing easier