Back to latest

Tech Videos

Sources AI Engineer All-In Podcast Andrej Karpathy Anthropic Apple Apple Developer AWS Events ByteByteGo Computerphile Cursor Dwarkesh Patel EO Fireship GitHub Google Cloud …

Sources

Watch First

DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux | Lex Fridman Podcast #501 · Lex Fridman In a sprawling, three-hour conversation, David Heinemeier Hansson (DHH) outlines his transition to 100% agent-driven development for his Hyprland-based Linux operating system, Omarchy Quattro, declaring that English is now the most beautiful and expressive programming language available to builders. It is a must-watch because DHH delivers a realistic, hype-free blueprint for how experienced builders can transition from hand-writing code to managing fleets of sub-agents with architectural integrity—specifically noting that while agents are brilliant for greenfield distro setups, establishing existing, mature web-app features still requires rigorous human-led architecture to avoid compiling “clanker slop”.


Highlights by Theme

Developer Tools & Platforms

In Slack Code Launch Event | Build Like Never Before, Slack showcases its new “Code Channels” as a dedicated multiplayer surface that lets engineers, product managers, and autonomous agents (like Devin) collaborate on PR diffs, UI mockups, and live testing in the open. On the Syntax channel, Wes Bos and Scott Tolinski walk through native web platform improvements in 8 New CSS, JS and HTML Features You Should Know, highlighting the ease of runtime relative color alpha transparency in CSS and the upcoming native promise.all keyed objects in JavaScript to bypass complex array indexes. In How We Got LLMs to Recommend Our Open Source Library — Christopher Burns, Inth on the AI Engineer channel, the founder of C15T shares how optimizing documentation pipelines using hand-written, brief llms.txt files and embedding twin markdown files directly into npm node packages yields up to 50% token savings for coding agents while completely bypassing costly HTML crawls.

AI & Machine Learning

On Google Cloud Tech, a comprehensive, hands-on JAX series including Build and optimize JAX training loops and Scale JAX models to multi-GPU systems provides engineers with deep implementation details on using ahead-of-time (AOT) compiling, optimizing memory layout with JAX’s shard_map for data parallelism, and managing StableHLO-based serializations to avoid first-call cold spikes in production serving. In The mathematics of AI uncertainty on Google DeepMind, Zoubin Ghahramani presents a compelling argument for moving beyond brute scale to innovate in Bayesian architectures that represent uncertainty explicitly, enabling models to perform continual learning without catastrophic forgetting in critical applications like AlphaFold and weather forecasting.

Hardware & Infrastructure

In AI Data Center Demand Isn’t Slowing: Southern CEO on Bloomberg Tech, Southern Company’s CEO discusses the logistical reality of the AI boom, showing how they manage massive infrastructure load agreements—such as OpenAI’s new 25-year Effingham County data center—using upfront collateral requirements and cancellation terms to protect the consumer grid. Meanwhile, on the hardware side, Meta Settles Federal Social Media Addiction Trial | Bloomberg Tech 8/26/2026 previews Nvidia’s earnings where investors are bracing for the Blackwell GPU rollout, with demand heavily outstripping supply constraints while the market watches for circular financing details.

Everything Else

In Why performant code matters (but gets widely ignored), with Casey Muratori on The Pragmatic Engineer, games industry veteran Casey Muratori argues that typical software runs 10x to 100x slower than it should, urging developers to learn to read assembly language to understand what compilers are actually delivering to the CPU. He debunks the lazy application of “premature optimization” rules, demonstrating that critical architectural bottlenecks (like serial dependency chains over network backplanes) must be solved at the design stage. Additionally, on Bloomberg Tech, the Meta Settlement Sets Blueprint for TikTok, YouTube details Meta’s landmark $16.7 billion settlement over social media addiction, highlighting a bipartisan coalition of attorneys general enforcing daily time caps and nighttime notification blocks for young users.


🎧 I can generate a podcast-style audio overview comparing DHH’s hyper-ambitious agentic workflow with Casey Muratori’s classic bare-metal performance principles if you’d like a brief to listen to on the go.

Search MacWorks

Enter at least two characters.