Search the archive

A SQL HEURISTIC: ORS ARE EXPENSIVE

OR clauses in SQL queries can be much slower than AND clauses due to how query planners use indexes. When using OR, the database might resort to less efficient strategies like merging separate filtered results or performing a full table scan. Rewriting queries to use AND clauses, often by using unions or restructuring the schema, can improve performance. Common schema issues that lead to needing OR clauses can be addressed by creating secondary tables or using an extension table pattern.

PARALLEL AND RECURSIVE ROUTE RENDERING

RSC routers use a unique approach to route rendering, performing it in parallel on the server and then recursively stitching the results together on the client to avoid waterfalls. They render a list of components in parallel, each potentially with a placeholder, and then serialize the output into a stream. On the client, a recursive component called `StackReader` uses the `Placeholder` component to traverse a stack-like data structure created from the stream, reassembling the nested component tree.

REACT VIEWTRANSITION: SMOOTH ANIMATIONS MADE SIMPLE

React is introducing a new experimental `ViewTransition` component that wraps the browser's View Transition API to create smooth, declarative animations between component states without animation libraries. The component automatically handles enter/exit animations, shared element transitions, and list filtering when used within React's concurrent features like `startTransition` or `useDeferredValue`. It allows CSS customization for different animation effects.

FRICTION IS NECESSARY FOR GROWTH

Overcoming friction is necessary for growth. Readily available AI could stifle creativity by removing the need to struggle and learn through mistakes. Excessive comfort reduces personal potential and can lead to stagnation of your skills.

TO AI OR NOT TO AI

An experiment was conducted over two weeks to build an app with full AI assistance. While the team found AI helpful in specific areas like searching, code snippets, and language tasks, they were frustrated with its limitations in providing context, maintaining code, and uncovering corner cases. Overall, the code became messy, control was lost, and the team returned to their traditional workflow.

INTRODUCING CLAUDE SONNET 4.5

Anthropic has released Claude Sonnet 4.5, a new frontier model with huge improvements in coding, computer usage, reasoning, and math. The model has state-of-the-art performance on benchmarks like SWE-bench Verified and OSWorld (for real-world software coding and computer task abilities). Along with the model, Anthropic has released Claude Code 2.0 and new features to the Claude apps.

BUY IT IN CHATGPT: INSTANT CHECKOUT AND THE AGENTIC COMMERCE PROTOCOL

OpenAI has launched Instant Checkout in ChatGPT, powered by the Agentic Commerce Protocol developed with Stripe. This allows users in the US to directly purchase items from participating merchants like Etsy and soon Shopify, within the chat interface. The Agentic Commerce Protocol, an open standard for AI commerce, allows AI agents, individuals, and businesses to collaborate on transactions while giving merchants control over payments and customer relationships.

GIVE YOUR AI EYES: INTRODUCING CHROME DEVTOOLS MCP

The Chrome DevTools MCP is a new tool that allows AI coding assistants to see and interact with a live Chrome browser through the Model Context Protocol, giving AI "eyes" to observe and debug web applications in real-time. The tool acts as a bridge between AI agents (like Cursor, Claude, and Gemini) and Chrome's DevTools capabilities, letting them navigate pages, inspect DOM elements, analyze performance, simulate user interactions, and debug issues based on actual browser feedback rather than coding blindly.

SUBTLETIES OF SQLITE INDEXES

This dev improved the performance of SQLite queries for his custom content feed by understanding how SQLite's query planner uses indexes. Composite indexes are better than multiple single-column indexes, and the order of columns in an index matters, following the "Left to right, no skipping, stops at the first range" rule. Also, the conditions in WHERE clauses for partial indexes must exactly match the corresponding query conditions.

SANDBOXING AGENTS AT THE KERNEL LEVEL

AI agents can be sandboxed at the kernel level in Linux to prevent unauthorized file access. Tracing the `open` syscall shows three points of failure where file access can be denied: permission checks, mount point redirections, and changing the root directory of the process. Combining mount namespaces and root changes offers control over the agent's filesystem view.

STOP USING .REVERSE().FIND(): MEET FINDLAST()

`Array.prototype.findLast()` and `Array.prototype.findLastIndex()` are cleaner, more performant alternatives to reversing arrays for searching from the end.

TEST DRIVEN DEVELOPMENT: BAD EXAMPLE

Test-driven development is sometimes a flawed theory supported by anecdotal evidence and doesn't always work in the real-world.

TUNING AIO IN POSTGRESQL 18

When tuning the new Asynchronous I/O (AIO) feature in PostgreSQL 18, users should generally stick with the default `io_method = worker` and increase the `io_workers` setting to around 25% of the system's cores.

ELIMINATING COLD STARTS 2: SHARD AND CONQUER

Cloudflare reduced cold starts for Workers by implementing a consistent hash ring to shard requests, improving memory efficiency and warm request rates.

→ Get the report ↗️

→ Tune into the on-demand SEO to GEO webinar [5] featuring leaders from Contentful, Vercel, and Apply Digital

APPLE BUILDS A CHATGPT-LIKE APP TO HELP TEST THE REVAMPED SIRI

Apple is currently using an internal ChatGPT-like iPhone app code-named Veritas to help test and prepare for a Siri overhaul planned for next year. The app is being used to quickly evaluate new features, including the ability to search through personal data and perform in-app actions. The internal tool marks a new phase in Apple's preparations for the Siri revamp, which could help reshape perceptions of its AI efforts.

META WANTS TO BECOME THE ‘BACKBONE' OF HUMANOID ROBOTS

Meta's chief technology officer, Andrew Bosworth, recently said in a meeting that he wants to focus on licensing Meta's software to other manufacturers instead of being a hardware manufacturer. Meta's new Superintelligence AI lab is working with the company's robotics group to build a 'world model' that can do software simulations to animate dexterous movements. Bosworth says that robotics will be Meta's next 'AR-sized' bet. The company has so far spent over $100 billion in cumulative investment into its augmented and virtual reality portfolio.

EVOLVING THE MULTI-USER SPACEPORT

SpaceX currently launches and lands its Falcon family of rockets every two days on average without interrupting the operations of other launch providers. Launch sites of the future will need to be fully operationalized like an airport, with multiple launches a day from multiple providers. SpaceX is working with federal regulators, the federal ranges, and industry partners to realize this vision. Its goals will advance the entire launch industry's ability to fly rapidly and reliably, ideally from multiple locations to increase resilience.

WHY TODAY'S HUMANOIDS WON'T LEARN DEXTERITY

Current projects developing humanoid robots will fail at teaching them dexterity because, in order to succeed, they need both the right data and for the robots to learn the right thing. Human dexterity relies on a rich sense of touch, and it often involves elbows, the front of the body, legs, and feet. Just using visual data is not enough. In the future, there will likely be a lot of robots with different forms for different specialized jobs that humans can do. They won't look human, but they will probably all still be called humanoid robots.

Schedule a demo to learn more

WHAT IS "GOOD TASTE" IN SOFTWARE ENGINEERING?

Taste is the ability to adopt the set of engineering values that fit the current project. It's different from skill - it's possible to be technically strong but have bad taste or be technically weak with good taste. Almost every decision in software engineering is a tradeoff. Often, you have to make tradeoffs between engineering values, such as resiliency, speed, readability, correctness, and flexibility, and that's where taste comes in. Bad taste is when your preferred values are not a good fit for the project.

SMALL DATA

Small Data is a movement away from Big Data toward much simpler and cheaper solutions. It is being driven by improvements in hardware and organizations that don't use that much data. Distributed systems are more overkill than ever. Developers should be considering what technologies they know well, are good enough, and are the least toilsome for the task at hand. Modern hardware will likely handle the workload just fine.

FAILING TO UNDERSTAND THE EXPONENTIAL, AGAIN

Many people look at the current state of AI, see the mistakes it makes, and then somehow jump to the conclusion that AI will never be able to do tasks at a human level. A few years ago, our current AI models were just science fiction. AI is improving at an exponential rate. It will be surprising if these improvements suddenly stop. Even a relatively conservative extrapolation of the trends suggests that by the end of 2027, models will frequently outperform experts on many tasks.

OPEN SOCIAL

Open source has clearly won, but this wasn't a foregone conclusion 35 years ago. There were many powerful forces that wanted open source to lose. We are now at a similar juncture with social apps. Projects like Bluesky's AT Protocol are changing the relationship between the user, the developer, and the project. What open source did for code, open social does for data.

the offer

ACCENTURE PLANS ON ‘EXITING' STAFF WHO CAN'T BE RESKILLED ON AI AMID RESTRUCTURING STRATEGY

Accenture's main strategy is upskilling, and it will exit people so it can get more of the skills it needs.

AMAZON'S AWS CEO CHIDES STAFF FOR SLOW PRODUCT ROLLOUTS

Amazon has been grappling with a reputation for slipping behind rivals in developing AI products.

THE AI BOOM NEEDS A MARKET FOR COMPUTE

Turning compute into a tradable commodity will be needed to fuel AI's next stage.

CALLED IT: THE ANTI-ADOBE COALITION IS HERE

Figma proved design doesn't need Illustrator, Descript proved editing doesn't need Premiere, and Omarchy and Framework proved developers don't need macOS.

Subscribe now.

U.S. MILITARY IS STRUGGLING TO DEPLOY AI WEAPONS

AVALANCHE STUDIOS NYC RETROSPECTIVE – AN AMBITIOUS COMPANY RUINED BY BAD DEVELOPMENT PRACTICES

Avalanche Studios was born out of ambition, but it failed to deliver on that ambition and closed a year ago.

↗️

APPLE'S ‘VERITAS' CHATBOT IS REPORTEDLY AN EMPLOYEE-ONLY TEST OF SIRI'S AI UPGRADES

An internal employee-only chatbot, codenamed “Veritas”, is speculated to be Apple's answer to ChatGPT, but it's unknown if or when it'll be released publicly. The company has struggled to get its AI efforts off the ground after a lukewarm reception to its Apple Intelligence features and multiple delays to Siri upgrades.

DEEPSEEK-V3.1-TERMINUS LAUNCHES WITH IMPROVED AGENTIC TOOL USE AND REDUCED LANGUAGE MIXING ERRORS

DeepSeek released V3.1-Terminus, which strengthens its Code Agent and Search Agent and addresses users' feedback about Chinese/English mix-ups. The model shows benchmark gains in agentic tool tasks (BrowseComp, SWE Verified, etc.) and offers both “chat” and “reasoner” modes for different use cases. Terminus remains open-weight under an MIT license, making it viable for customization and deployment both via API and local environments.

HEAD OF CHATGPT ADDRESSES CONTROVERSIAL MODEL ROUTING

OpenAI confirmed it's testing a safety system that automatically switches users from GPT-4o to reasoning models or GPT-5 when conversations touch on "sensitive and emotional topics." Users have complained that the routing triggers too frequently in harmless situations, pulling them away from GPT-4o, which many prefer for personal conversations.

THE AI VILLAGE IN NUMBERS

OpenAI's models lead on various stylistic measures like lexical diversity, positive sentiment, and formality of language, but Anthropic's models lead in goal-directed behavior. While OpenAI's models vary in which stylistic measures they excel at, Anthropic's models all perform well on agentic behavior. OpenAI offers the brightest, most cheerful, and most eloquent model. Anthropic's models tend to excel at getting stuff done.

OPENAI RELEASES WORKPLACE USAGE REPORT SHOWING RAPID CHATGPT ADOPTION

28% of US workers now use ChatGPT at work, with adoption highest among highly educated white collar workers in tech, finance, and manufacturing. It's vague exactly how the tool is being used - OpenAI is primarily for writing, research, programming, and data analysis - but technical teams have emerged as the clear power users.

scale. Save your spot

MODULAR MANIFOLDS - THINKING MACHINES LAB

Normally, neural networks are trained by letting their weights move freely in any mathematical direction. A new method that restricts the weights to specific curved surfaces makes training more stable and predictable. This "manifold Muon" optimizer forces weight matrices to maintain consistent properties (like not stretching or shrinking data too dramatically) by constraining them to geometric shapes called manifolds. While promising, the experiments required computational overhead, which limits real-world adoption.

WE REVERSE-ENGINEERED FLASH ATTENTION 4

Flash Attention 4 is a brand new, highly optimized CUDA kernel that speeds up the attention calculations in transformers, which are the main mathematical operations that bottleneck AI models like ChatGPT. However, the improvements from FA4 are not from clever new math but from improvements in how an incredibly complex async pipeline splits computations across 32-thread ‘warps'.

WHAT PROBLEM IS DSPY SOLVING WITH WHAT ASSUMPTIONS

DSPY's programming model is a good base, but it has issues with defaults and missing affordances. This post identifies these issues and suggests some changes. Adopting these changes would sharpen DSPy's design without fighting it. It would keep the ergonomics that makes teams productive while aligning the optimization layer with the realities of fixed budgets and production constraints.

ALIBABA BETS BIG ON AI WITH NVIDIA TIE-UP, NEW DATA CENTER PLANS

Alibaba has partnered with Nvidia to enhance its cloud platform with advanced AI capabilities, focusing on robotics and autonomous driving. It plans to expand global data centers, opening facilities in Brazil, France, and the Netherlands to meet increasing AI infrastructure demand. The company also launched Qwen3-Max and Qwen3-Omni AI models, emphasizing improved performance in code generation and virtual reality applications.

EXPANDING OUR DATA ENGINE FOR PHYSICAL AI

Scale is expanding its Data Engine for Physical AI, providing robotics companies with high-quality, annotated datasets needed for training foundation models in physical environments. It addresses the robotics data gap by ensuring datasets are abundant, diverse, and enriched, collecting data through dedicated robots and human demonstrations. This aims to accelerate the development of reliable AI systems capable of handling real-world complexity in various applications.