Search the archive

WORDPRESS DROPS TO 41.5% MARKET SHARE, BUT THREE DATASETS TELL DIFFERENT STORIES ABOUT WHAT'S GOING ON

WordPress now powers 41.5% of the web, down from a peak of 43.6% in early 2025.

AGENTIC LOOPS

A look at various strategies to run agentic loops.

APPLE PLANS FIVE NEW IPHONES THROUGH 2027, EYES CHINESE-MADE CHIPS AMID FOLDABLE PUSH

Apple has instructed suppliers to prepare to produce about 10 million foldable iPhones this year.

DOES A URL IN A PROMPT STEER AN LLM'S OUTPUT TOWARD ITS CONTENT?

When a URL and its content make it into the training data, the bare URL works as a key into the weights

SOME NEW AGENTIC PATTERNS

A look at how Prime Radiant uses agents.

Try Bob for free

BYTEDANCE SET TO LAUNCH SEEDANCE 2.5 WITH 3-MINUTE AI VIDEO OUTPUT

ByteDance is rumored to be launching its Dreamina Seedance 2.5 model on July 9. Availability is expected across Dreamina, CapCut, and other ByteDance partner platforms. The new model can output 180-second-long videos, but it is unclear whether it can keep character identity, motion, camera logic, and prompt intent stable.

ALIBABA REPORTEDLY RESTRICTED CLAUDE CODE

Alibaba reportedly planned to prohibit employee use of Claude Code from July 10 after classifying it as high-risk software. Employees were instead directed toward Alibaba's Qoder tool amid Anthropic's efforts to prevent unauthorized access and model distillation.

OPENAI MIGHT BE PREPARING GPT-5.6 FOR NEXT WEEK'S RELEASE

OpenAI has moved GPT-5.6 into a narrow preview, splitting it into three tiers: Sol, Terra, and Luna. A notable change includes a new reasoning-effort control slider and an "ultra" mode for handling complex tasks. The release has implications for Codex users, amid Anthropic's Fable 5 changes, while broad access depends on US government review approvals.

CLOSING THE VERIFICATION LOOP

Agents have made building cheap, but the cost just moved to verifying whether things actually work. The verification loop is the distance between a claim and its proof. Most teams close this loop with humans, but agents are starting to outpace humans. Compound Engineering is a plugin with skills for coding agents. This post discusses how /ce-dogfood, a skill within Compound Engineering, closes the loop and the persona strategy that gives it eyes.

CLOUDED JUDGEMENT 7.3.26 - THE END OF COMPUTE SCARCITY? NOT SO FAST

Meta and SpaceX selling compute capacity could mean that there's excess compute capacity that two of the largest buyers no longer need. This may result in downward capex revisions for the hyperscalers. It's possible demand simply isn't there for what's being built with AI. However, this is unlikely, as anyone willing to sell capacity finds buyers immediately.

UNDERSTANDING THE DYNAMICS OF THE AI ECOSYSTEM WITH PACE LAYERS

Pace Layers is a framework for organizing fields and categories by how fast they change. Each layer is a differently paced component that is functionally different from the others and operates somewhat independently, but each layer influences and responds to the layers closest to it in a way that makes the whole system resilient. This post applies the Pace Layers framework to the AI ecosystem to help readers process its dynamics.

JAMESOB'S GUIDE TO RUNNING SOTA LLMS LOCALLY

This guide discusses the hardware used to run SOTA AI models locally and how to run speech-to-text (STT) locally. It also provides a ready-to-run configuration for running models within Docker containers. A budget of $2,000 can get you a setup that runs Qwen and good STT. $40,000 can get you a machine that can run almost-Opus-level models.

A FIELD GUIDE TO FABLE: FINDING YOUR UNKNOWNS

The difference between the map and the territory is the unknown. When Claude runs into an unknown, it needs to make a decision based on its best guess. The more work that is done, the more unknowns Claude might run into. This post discusses how to use Claude to help discover unknowns. Discovering your unknowns before starting a project is a cheap way to find out what you didn't know before it gets expensive to fix.

OPEN SOURCE AI GAP MAP

The Gap Map maps out the open-source AI stack so people can collectively build what's missing. The current open source AI stack is seriously robust, but fragmented, duplicative, and hard to see as a coherent whole. The Gap Map helps people identify where to build new, where to invest in capability, and where to open up the tools. It is an invitation for greater collaboration and coordination across the space.

OWN THE LOOP: A FIELD GUIDE TO AGENT HARNESSES

As coding models commoditize, the real differentiator is the harness: the control loop that manages tools, workflows, orchestration, and model routing. Vendor-native harnesses offer the best performance today, but owning a portable, model-agnostic loop may become the more durable advantage as models evolve and costs fall.

A BRIEF HISTORY OF DISTILLATION IN AI

Distillation started as a way to compress large models into smaller, cheaper ones, but has evolved into a core post-training technique for transferring instruction-following and reasoning abilities from frontier models. Methods used by DeepSeek, Qwen, and GLM now sit at the center of both open model development and ongoing disputes over whether training on proprietary model outputs constitutes unauthorized copying.

LET FABLE USE ITS OWN JUDGEMENT RATHER THAN DICTATING HOW IT SHOULD WORK

It's better to tell Fable to use its own judgment when deciding to write tests and when choosing other models for smaller tasks.

LEANSTRAL

Leanstral is an open-source 119B-parameter theorem-proving and code-verification agent built on Mistral's general-purpose coding framework.

GOOGLE TESTS NEW GEMINI INBOX SECTION FOR WORKSPACE TRIAGE

Google is testing a new dedicated inbox section within the Gemini app for Business and Workspace users.

APPLE REPORTEDLY TESTING NEW IPAD PRO MODELS AND REDESIGNED ENTRY-LEVEL MACBOOK PRO FOR 2027

Apple is reportedly testing four new 11-inch and 13-inch iPad Pro models for a spring 2027 launch, focusing mainly on internal upgrades such as faster chips and potentially vapor chamber cooling, while also planning a redesigned 14-inch entry-level MacBook Pro that could arrive as early as the first half of 2027. Bloomberg also says Apple is accelerating the launch of its M7 chip to better handle AI workloads, following the upcoming M6 generation. Although memory and chip shortages could still affect the roadmap, these devices are expected to contribute to what could be Apple's biggest product year yet in 2027.

MIDJOURNEY WANTS HOLLYWOOD STUDIOS TO REVEAL THE DETAILS OF THEIR AI USAGE

Midjourney is pushing to force Disney, Universal, and Warner Bros. to disclose their own internal AI usage during discovery in the ongoing copyright lawsuits against it. The studios currently only must share generative AI documentation tied to "consumer-facing" content, a limit Midjourney's new filing seeks to remove, arguing it lets studios hide evidence of similar internal AI practices while cherry-picking material against Midjourney. The company also wants studios to disclose all Midjourney prompts and outputs used, not just those tied to allegedly infringing images, while the studios' attorney has dismissed the request as a "fishing expedition."

APPLE DESIGN AWARD WINNER ACQUIRED BY APPLE FOR NEW SWIFT TOOLS

Apple acquired Rabbit 3 Times, Inc, maker of the visual Swift development tool Play, months after giving Play an Apple Design Award for innovation. The deal, reported to the EU under the Digital Markets Act, reportedly involved acquiring assets and the option to hire staff rather than a full acquisition. Following the deal, Rabbit 3 Times discontinued Play for iPhone and Mac, made its paid Play-to-Xcode export service free, and later shut down its website.

YOU DESIGN IT. THEN WHAT? A CLEAR MAP OF THE FIGMA-TO-CODE AI MESS

AI design-to-code workflows are best understood as a four-layer stack: MCP connects AI to tools like Figma, markdown files capture design rules and context, skills automate repeatable tasks, and mapping (such as Code Connect) links Figma components to real code components—each layer solving a different problem while still requiring increasingly more engineering effort. For most designers and small teams, a clean Figma file, an MCP connection, and a simple Markdown guide are enough. Only mature design systems benefit from the full stack, as maintaining design-to-code synchronization is expensive and developer-dependent. AI dramatically speeds up turning designs into code, but it doesn't eliminate the need for structured design systems, human judgment, or developers to build and maintain production-ready software.

YOUR INTERFACE HAS A TONE. AND SOMETIMES IT BLAMES YOU

Many digital products are designed around an unrealistic "ideal user" who patiently adapts to confusing interfaces, causing people to blame themselves instead of poor design when something goes wrong. Inclusive UX writing—using clear, supportive, non-judgmental language—reduces cognitive burden for everyone, especially people with disabilities, anxiety, ADHD, or language barriers, reflecting the curb-cut effect, where designing for those at the margins improves experiences for all. Every piece of interface copy is a design choice, and thoughtful language can make users feel guided rather than judged.

RETHINKING FIGMA IN AN AI WORLD

At Config 2026, Figma expanded its canvas with code layers, motion, and shader tools to keep design relevant as AI shifts workflows toward code-native, agentic product building. This pressures Figma's seat-based business model, since AI lets engineers and teams generate or inspect software directly without a shared design file. Whether Figma stays central now depends on future workflow behavior, not conference hype, as rivals like Anthropic's Claude Code and Claude Design push teams to skip design handoffs entirely.

REFERENCE BOARD

Reference is a private moodboard app for iPhone, iPad, and Mac. Save inspiration, organize moodboards, find similar images, and sync creative references.

DESIGN.MD COLLECTION FOR AI CODING AGENTS

Browse independent design system analyses from popular websites. Drop one into your project as a design reference for your coding agent.

YOUR ALL-IN-ONE AI WORKSPACE

Genspark is your all-in-one AI workspace. Slides, docs, images, video, code, and design — all in one place.

WHY A BITTEN ‘T' FROM 1889 WAS THE INSPIRATION FOR FATYPE'S STATES

Inspired by the quirky imperfections of late-19th-century American sans serif typefaces, Anton designed States, a new type family that embraces the irregularities of historic metal type while balancing personality with readability. The family includes States Grotesque and States Rounded, linked by a variable Pressure axis that lets users smoothly adjust between sharp and softened letterforms, with numerous stylistic alternates inspired by vintage printing. To accompany the release, the foundry also produced its first printed type specimen in years, showcasing the typeface across a range of sizes.

THE WORLD'S OLDEST FOOTBALL CLUB JUST MODERNISED ITS LOGO, AND FANS ARE FURIOUS

The world's oldest football club, Sheffield FC, has unveiled a radically simplified circular logo for the 2026–27 season, replacing its historic shield, figures, and traditional design elements with a modern roundel intended to reflect its unique heritage and improve digital versatility. While the club says the redesign marks a new chapter, many fans have criticised it as generic, poorly executed, and disconnected from the club's 169-year identity, arguing that modernization could have preserved more of its historic character.

AI MADE DESIGN FASTER, BUT DID IT MAKE IT LESS INTENTIONAL?

AI has sped up design production, but polished outputs often mask missing edge cases, technical constraints, and system-level coherence. As generation gets easier, evaluating whether solutions solve the right problem becomes the harder, more valuable task. Designers who thrive will slow down key decisions, review work systemically, and treat AI as a collaborator rather than a final authority.

THIS TINY MACBOOK ACCESSORY ADDS CUSTOMIZABLE SHORTCUTS FOR MEETINGS AND PRODUCTIVITY

Dune is a USB-C accessory for MacBooks that adds three AI-powered, app-aware programmable buttons to simplify shortcuts and automate common tasks.

APPLE'S LIQUID GLASS IS HAVING PROBLEMS… AGAIN

A CONTEMPORARY INDIAN ILLUSTRATOR CREATING IMAGES FILLED WITH STORYBOOK LIGHT, STRUCTURE, AND EMOTION

Alfred Fernandes is a Goa-based illustrator and concept artist known for storybook-like images blending visual development, environment design, and narrative-driven art.

HUNTING A 16-YEAR-OLD SQLITE BUG WITH TLA+: IS DQLITE AFFECTED?

SQLite recently addressed a long-standing bug related to the Write Ahead Log (WAL) that could cause database corruption, present in the system for 16 years. The dqlite team investigated whether this bug affects their implementation by modeling SQLite's behavior in TLA+ and comparing it with how dqlite interacts with SQLite and concluded that dqlite is not affected due to its stricter locking mechanisms that prevent simultaneous writes and checkpoints.

AGENTIC TEST PROCESSES, LLM BENCHMARKS, AND OTHER NOTES ON AGENTIC CODING FROM GALAPAGOS ISLAND

The exploration of coding agents and their testing capabilities shows a disconnect between the efficiency of AI in automating processes and the reliability of its outputs, especially in coding and debugging tasks. While AI models can speed up workflows and increase code quality through good testing methods, they often struggle with generating reliable tests and identifying the true source of bugs.

POSTGRESQL AND THE OOM KILLER: WHY WE USE STRICT MEMORY OVERCOMMIT

Strict memory overcommit in PostgreSQL is necessary for preventing catastrophic out-of-memory (OOM) kills, as it allows the database to handle memory allocation failures gracefully without impacting all connections. A kernel bug in Ubicloud's product temporarily disrupted this configuration, showing the importance of careful management of memory limits to guarantee system stability while running PostgreSQL in production environments.

BETTER MODELS: WORSE TOOLS

Recent iterations of Anthropic's models, specifically Opus 4.8 and Sonnet 5, show a troubling trend of producing malformed tool calls when interacting with Pi's editing tool, as they frequently append extraneous fields that do not conform to the expected schema. This regression seems to be linked to the models' training environments, which appear to favor a more forgiving approach to tool calls.

WRITE CODE, NOT SPECS

Good code should be self-documenting, allowing for an understanding of its functionality without extensive specifications. Instead of maintaining both code and specifications, focusing on writing code that establishes patterns and relying on tests to validate functionality allows for greater trust and efficiency in development.

SOME NEW AGENTIC PATTERNS

At Prime Radiant, new "agentic" patterns have been implemented to enhance collaboration between agents and human implementers, resulting in improved tools and workflows. Notable examples include an agent named Nora, which acts as a team member rather than just an assistant, and a framework allowing multiple agents to collaboratively address tasks, leading to efficient problem-solving and feature development.

PLANNOTATOR

Plannotator is a browser-based tool that enables users to annotate and review plans, specifications, and code changes from various AI coding agents. It integrates with multiple platforms, supports real-time collaboration, and allows for easy sharing of annotated content with team members.

BRAMBLE

Bramble is a local-first, open-source password manager that encrypts and stores user data on personal devices without relying on cloud servers, allowing for secure peer-to-peer syncing and various authentication methods, including biometric options and passkeys.

STATE OF CLI CODING AGENTS, MID-2026

As of mid-2026, the terminal has become a favored platform for coding agents, with 35 actively maintained CLI agents available. The main players like Claude Code, Codex CLI, and Omp have come up as leading options, each having features such as memory capabilities, editing precision, and orchestration.

PERFORMANCE PER DOLLAR IS GETTING FASTER AND CHEAPER

The release of GLM-5.2 shows advancements in inference performance, with 2626 tok/s/node on AMD MI355X at more than twice the cost-efficiency compared to similar NVIDIA setups. Despite historical challenges with AMD's software support, improvements in model optimization are narrowing the gap.

JAMESOB'S GUIDE TO RUNNING SOTA LLMS LOCALLY

The guide shows detailed instructions and recommendations for setting up a state-of-the-art local machine learning system using specific hardware configurations to run advanced LLMs properly.

WHAT ORMS HAVE TAUGHT ME: JUST LEARN SQL

ORMs can provide convenience in working with SQL but often complicate data retrieval and management due to issues like attribute creep, inefficient queries, and dual schemas.