Search the archive

WHO'S AFRAID OF CHINESE MODELS?

Chinese models are less of an economic threat than restrictive US AI policies. America should expand access to powerful models and support open innovation.

WHAT MR. JAMBO AND LEVI'S CAN TEACH US ABOUT LISTENING TO CUSTOMERS

Leaders often reshape customer feedback to confirm existing beliefs. Teams should test their assumptions and consider that customers may mean exactly what they say.

THE DENOMINATOR THAT MATTERS

Successful AI adoption requires technical understanding, problem knowledge, evolving practices, and trust. Without a credible social contract, resistance is rational.

THE POWER OF “WHY?”

Requirements often prescribe solutions instead of identifying real needs. Asking “why” and testing small increments helps teams preserve options and find simpler answers.

THIS IS HOW TO LIVE A GOOD LIFE

A good life is built through repeated action, not insight alone. We become better by practicing virtue, pursuing meaningful goals, and investing in relationships.

THE RUDE DUCK: WHAT ARE YOU AVOIDING?

Fear often hides behind reasonable excuses.

AI'S NEXT WINNERS WON'T JUST BE THE FRONTIER LABS

AI's value is spreading beyond frontier labs.

paying attention. See how TLDR sponsorship works.

META IN TALKS TO LEASE COMPUTING POWER TO ANTHROPIC IN POTENTIAL $10 BILLION DEAL

Anthropic proposed a computing deal with Meta in June that could be worth as much as $10 billion over two years. Meta is considering the deal, which would involve monthly payments and the option to opt out of any agreement early. The talks show how demand for more computing power is still high. The deal could open up a new line of business for Meta and provide Meta with a new revenue stream until demand for its own AI services catches up.

SPACEX IN TALKS TO PROVIDE COMPUTING POWER FOR PENTAGON'S AI PUSH

SpaceX is in talks to provide the Defense Department with computing capacity at a cost of up to several billion dollars. Some national security officials have raised concerns that the Pentagon is becoming too reliant on Elon Musk's services. Pentagon officials have said they want to reduce reliance on individual tech companies. The Pentagon is seeking $30 billion in funding for an initiative that will focus on securing high-end AI chips.

INDIA'S FIRST PRIVATELY-DEVELOPED ROCKET REACHES ORBIT ON DRAMATIC DEBUT LAUNCH

Skyroot Aerospace's Vikram-1 rocket, India's first fully commercial satellite launcher, launched into a 280-mile-high orbit from an island spaceport in the Bay of Bengal on Saturday. The launch was delayed for more than half an hour to resolve a last-minute technical problem, but the flight was deemed a grand success. The Vikram-1 is India's first private orbital rocket to reach orbit on its very first attempt. A second Vikram-1 launch could happen before the end of the year.

CHINA JOINS RUSH TO RETHINK THE SMARTPHONE FOR THE AI ERA

Several companies are now racing to develop devices with built-in AI agents. ZTE Corp. has unveiled a lineup of co-designed smartphones with built-in AI services. The NaviX Ultra, which it calls the world's first agentic smartphone, pulls up ByteDance's AI agent Doubao with a voice command or the press of a button. It comes in black, pink, white, and blue.

one click. Get started →

HOW ANTHROPIC RUNS LARGE-SCALE CODE MIGRATIONS WITH CLAUDE CODE

Code migrations were typically multi-year endeavors until recently. Claude Code's capabilities change the math for these types of projects. This post details the six-step process Anthropic uses to perform code migrations. The core principle is to fix the process that produces the code rather than fixing the code itself.

SCROLL-WORLD

scroll-world is a skill that turns any brand into a scrollable 3D world. It works with Claude Code, Codex, and any SKILL.md-compatible agent. The skill builds immersive, control-scrubbed 'fly through the world' landing pages for any industry or brand. A video example is available in the repository.

MAYBE INTELLIGENCE AIN'T ALL THAT

AI chatbots are now smarter and more capable than a lot of people, but there has yet to be an explosion or take-off in usage. Coming up with ideas was never the hard part. Most problems can't be worked out at a desk by thinking. Reality stays bound by how fast a thing can be built and whether it holds.

THESE AI-NATIVE COMPANIES HAVE TINY STAFFS AND FEWER BOSSES

The newest generation of AI-infused companies offer a vision of how work could soon be structured in other American corporations. They have fewer workers, more on-staff engineers, and a flatter structure where everyone is a player-coach instead of strictly overseeing teams. The structure is a proving ground for the philosophy that corporate giants are trying to channel through reorganizations, de-layering, job cuts, and investment into AI.

APPLE, NVIDIA VIE FOR TITLE OF WORLD'S MOST VALUABLE COMPANY

Apple briefly became the world's most valuable company on Friday.

AN AWS BILLING BUG SENT USERS ESTIMATED CHARGES OF UP TO $2.5 TRILLION

Amazon has confirmed a unit pricing error and is recomputing all estimates.

ARE THE LLM WARS THE DATABASE WARS?

The winners of the database wars weren't the ones everyone was arguing about.

DOMAIN-SPECIFIC HARNESSES

Different workloads perform better with certain combinations of models and harnesses.

QWEN3.8 HAS LAUNCHED

The massive 2.4T parameter model is also going open weight soon.

AI MANIA IS EVISCERATING GLOBAL DECISIONMAKING

The people in charge either have no plan or see no path forward other than keeping their heads down.

QWEN3.8 IS GOING OPEN-WEIGHT

Alibaba announced Qwen3.8, a 2.4-trillion-parameter model slated for an open-weight release. A preview version is available through Alibaba's Token Plan, Qoder, and QoderWork.

APPLE SENDS LEGAL LETTERS TO DOZENS OF OPENAI EMPLOYEES

Apple has reportedly sent legal letters to dozens of former Apple employees now working at OpenAI. The letters instruct the employees to preserve potentially relevant documents and communications related to Apple's trade secret lawsuit against OpenAI. Apple claims that OpenAI recruited key engineers and benefited from proprietary designs, manufacturing processes, and other trade secrets. More than 400 former Apple employees now work at OpenAI. OpenAI has denied the allegations, saying it is not aware of any evidence that the complaint has merit.

FABLE 5 VS. GPT-5.6 SOL ON AN NP-HARD PROBLEM: DOES /GOAL HELP?

Claude Code and Codex both have a /goal option, but their implementations are fundamentally different. Claude Code implements /goal as a session-scoped Stop hook, so it can catch an early exit, but it doesn't know whether another ten million solver iterations are worthwhile. Codex treats a goal as a persisted thread state. It sees the files and tools, but effectively grades its own work. Using /goal can be a bad default as it can amplify bad decisions as much as good ones.

HOW NETFLIX BUILT ITS LLM SERVING STACK

Netflix details how it deployed and operated LLM inference within its existing production infrastructure. The article covers engine selection, model packaging, API design, deployment strategy, output constraints, and trade-offs that emerged under real workloads.

DEMIS HASSABIS AND GOOGLE'S AI COMMITMENTS

This post examines Demis Hassabis's framework for frontier AI alongside criticism of Google's military agreements. It also covers Alex Turner's resignation after unsuccessfully opposing broad government use of Google's models, including for autonomous weapons.

works

DIFFUSING BLAME: DALE-CONSTRAINED LEARNING WITHOUT WEIGHT TRANSPORT

Sakana AI's "Diffusing Blame" is a neural network technique that enforces Dale's principle while enabling effective learning in image classification and reinforcement learning tasks. Their approach, based on Error Diffusion, uses non-negative weight matrices to maintain separate excitatory and inhibitory streams, eliminating the need for weight transport. Results show strong alignment with true gradients, indicating a step toward biologically plausible learning rules.

KIMI CODE CLI

Kimi Code CLI is an AI coding agent that runs in the terminal. It can read and edit code, run shell commands, search files, fetch web pages, and choose the next step based on feedback. The CLI works with Moonshot AI's Kimi models and can be configured to use other compatible providers. Its compatibilities can be extended through agent skills, hooks, sub-agents, and MCPs.

CLAUDE FABLE 5 WILL BE INCLUDED IN ALL MAX AND TEAM PREMIUM PLANS

Claude Fable 5 will be included in all Max and Team Premium plans at 50% of limits starting July 20. Pro and Team Standard users will continue to have access to Fable via usage credits. They will receive a one-time $100 credit. Fable is being rolled out in stages as demand for the model has been challenging to predict.

ALIBABA OPEN-SOURCES ITS AI CHIP SOFTWARE STACK AT WAIC, TARGETING NVIDIA'S CUDA LOCK-IN

Alibaba is open sourcing the software stack for its Zhenwu series of AI chips. The move will lower migration barriers for developers currently locked into Nvidia's CUDA ecosystem. If China wants AI independence, it needs to break the CUDA lock-in at the software layer, not just build alternative chips. The Pentagon recently added Alibaba to its Chinese military companies blacklist. Open sourcing SAIL will make its ecosystem stickier and harder for any single government to shut down.

KIMI K3 HAS RECEIVED FAR MORE LOVE THAN EXPECTED

GOOGLE PREPARING SKILLS AND GEMINI LIVE FOR WEB ROLLOUT

Google plans to introduce Gemini Live to desktop and web, extending real-time voice capabilities beyond mobile.

I BURNED ALL MY TOKENS RESEARCHING HOW TO SAVE TOKENS

Use cheap models to find information, accurate models to verify it, and then use deep research last.

A CHINESE AI STARTUP IS ABOUT TO HIT $1BN IN SALES WHILE GIVING ITS BEST MODELS AWAY FOR FREE

A large share of Z.ai's income comes from on-premises deployments for state-owned enterprises and financial institutions, as well as its fast-growing cloud business.

MOONSHOT AI PLANS HONG KONG IPO AFTER KIMI K3 MODEL DEBUT

Moonshot AI plans to list on the Hong Kong Stock Exchange within six months.

WHY THE FIRST GPU FINANCIERS ARE TURNING TO INFERENCE CHIPS IN A $400 MILLION DEAL

General Compute, an AI inference cloud startup, is using inference-specific chips as collateral for a $400 million loan.

OPENAI TWEAKS CHAT ACCESS IN THE CHATGPT APP FOR MAC

OpenAI has updated its redesigned ChatGPT app for Mac after widespread criticism of its confusing interface, restoring easy access to chat history and Projects. The app now features a clear toggle between Chat and Work, while a sidebar menu lets users switch between ChatGPT and Codex, making the standard chat experience much easier to find again. Although the update fixes the biggest usability complaint, the app still relies on a larger Electron-based framework and has a more complex interface than before, leading some to question OpenAI's focus on integrating new AI tools over maintaining a polished Mac experience.

GOOGLE VIDS NOW LETS YOU STAR IN YOUR OWN AI VIDEOS

Google Vids now lets users create personalized AI avatars from a selfie and voice recording, tied to their Google account and invisibly watermarked with SynthID. The update also brings the Gemini Omni model to Vids, enabling video creation from prompts and reference images, background/lighting fixes, and step-by-step edits without restarting. These additions expand Vids from a workplace presentation tool into a broader video platform, positioning it against avatar competitors like HeyGen and Synthesia. Access is restricted to users 18+ in select regions.

BRING CODE-BACKED SCREENS ONTO THE CANVAS WITH VARIABLES ATTACHED

Figma now binds colors, type, and spacing to existing file variables when code-backed screens are brought onto the canvas, rather than importing them as hardcoded values. This applies to Make preview copies, code-to-canvas generation via the Figma MCP server, and live web pages captured through the Chrome extension. Frames also arrive with auto layout more often, letting edits resize and reflow so teams can refine designs instead of rebuilding them.

META'S "REGRET" OVER ITS SCRAPPED AI TOOL IS LIKE A BURGLAR APOLOGISING FOR THE MESS

Meta briefly introduced a feature in its new Muse Image AI tool that allowed users to generate and edit images using photos from any public Instagram account without prior consent, with the feature enabled by default. After widespread criticism from creative professionals, unions, talent agencies, and privacy advocates, the company removed the feature within three days, acknowledging that it had "missed the mark." The incident highlights a broader concern about AI companies launching opt-out features that treat publicly shared content as available for AI use unless users explicitly disable it. It also underscores the importance of consent, creator rights, and vigilance, as platforms continue to push AI capabilities that can affect photographers, designers, illustrators, and other creative professionals.

WHY IS AI BAD AT DESIGN?

AI struggles with design because, unlike code, design has no automatic way to verify if an output is correct, forcing reliance on slow, subjective human judgment. This subjectivity pulls AI toward generic, "average" outputs, worsened by scarce design training data and models generating interfaces blindly, without truly seeing what they create. While vision-based feedback loops and better models may narrow some gaps, the subjective nature of taste remains a structural limit, meaning humans will still be essential for genuine design judgment and originality.

THE AGENTIC CREATIVE STUDIO

Miora reads your canvas, remembers your style, and iterates with you. Not a one-and-done tool — a creative partner that gets smarter every session.

AI 3D MODEL GENERATOR

Create 3D models, characters, game assets, and animations with V2Fun's AI 3D model generator.

ON-BRAND ADS, GENERATED IN SECONDS

Paste a public domain, and Branda pulls the logo, colors, style, and products, then generates 4–6 ready-to-ship ads: three for the company and up to three for its products.