Search the archive

CODING IS A META-TASK

Coding is fundamentally a structured type of problem-solving, so when models get better at coding, they get better at solving all other kinds of problems too.

I AM BUILDING A CLOUD

exe.dev addresses the VM resource isolation problem: instead of provisioning individual VMs, users get CPU and memory and run the VMs they want.

Explore Crusoe Managed Inference

GPT 5.5

OpenAI released GPT-5.5 with improved agentic reasoning, tool use, and efficiency, matching prior latency while increasing performance across coding and knowledge tasks.

DEEPSEEK UNVEILS FLAGSHIP AI MODEL A YEAR AFTER BREAKTHROUGH

DeepSeek has rolled out new preview versions of its new flagship AI model. The V4 Flash and V4 Pro series comes with architecture upgrades and optimization improvements. It has a 1-million-token context window. While the service capacity for the V4 Pro series is extremely limited due to a computing crunch, DeepSeek expects pricing for the model to drop once Huawei releases its Ascend 950-powered computing clusters.

TENCENT, ALIBABA TO BACK DEEPSEEK AT $20B+ VALUATION

DeepSeek is in talks for its first funding round at a $20 billion valuation, with Tencent and Alibaba interested. Tencent is seeking a 20% stake, but DeepSeek doesn't want to lose that much control. The valuation surged from $10 billion to $20 billion in days, illustrating significant investor interest.

ANTHROPIC JUST OVERTOOK OPENAI WITH $1 TRILLION VALUATION

Anthropic reached a $1 trillion valuation on Forge Global, surpassing OpenAI's $880 billion. The spike in Anthropic's value stems from scarce available shares and growing demand due to partnerships and increased use of its Claude Code tool. Current investor interest highlights Anthropic's position as a leading AI opportunity.

TRAINING FOR ACCURACY IN SEARCH LLMS

Perplexity detailed a staged post-training approach that teaches models tool use, evidence gathering, and structured evaluation to improve accuracy in search-based AI systems.

AGENTICS: AI ENABLEMENT REQUIRES MANAGED AGENT RUNTIMES

Enterprises need managed agent runtimes for effective AI integration, avoiding the complexities of manual configuration. Current solutions like Claude Code are unintuitive, leading to inefficient setups and potential security risks. Companies like Ramp and Stripe have developed custom solutions, but managed services are scarce, creating opportunities for new entrants in the market.

Read how Cursor cut costs 20x and improved agent reasoning

AN UPDATE ON RECENT CLAUDE CODE QUALITY REPORTS

OPENAI PRIVACY FILTER MODEL

OpenAI released a lightweight open-weight model for detecting and redacting PII in text, designed for fast, local, context-aware privacy filtering workflows.

EXPERT UPCYCLING

Amazon researchers released a method to expand Mixture-of-Experts models mid-training by duplicating and specializing experts, improving performance without increasing inference cost.

AI CODING FIRM COGNITION IN FUNDING TALKS AT $25 BILLION VALUE

Cognition AI is in early talks to raise a round of funding that would more than double its valuation to $25 billion. The talks are ongoing and the terms could change. Cognition uses AI to streamline the process of writing and debugging code, with a focus on selling to businesses. Its flagship product, Devin, is being used by companies like Anduril and Microsoft.

ORACLE'S DELUGE OF AI DEBT PUSHES WALL STREET TO THE LIMIT

Oracle's $300 billion megadeal with OpenAI is testing the limit of Wall Street's appetite for debt tied to the datacenter boom. Banks have struggled for months to spread the risk of the billions of dollars in loans they made to build data centers leased to Oracle in Texas and Wisconsin. Bank balance sheets are now clogged, constraining the financing prospects of future projects tied to Oracle and OpenAI. Silicon Valley needs access to debt to meet its goals for AI-related spending, but so far, Wall Street is largely giving a blank check for the AI ambitions of the most creditworthy tech companies.

AGENTS CAN'T CHOOSE BETWEEN STRUCTURE AND FLEXIBILITY

Deciding between Python and Markdown for agent specification involves balancing structure with flexibility. Code-maximalism enforces reliability but lacks adaptability, while Markdown-maximalism allows for creativity but risks errors. Successful agent architectures use a hybrid approach, leveraging Markdown for intent and code for structure to combine flexibility with reliable execution.

Here's why.

AI SUMMARIES IN GMAIL

Google added AI Overviews to Gmail for Workspace users, enabling natural language queries that return summarized answers across emails without opening threads.

MICROSOFT TO INVEST $1.8B IN AUSTRALIA TO EXPAND AI, CLOUD, AND DIGITAL INFRASTRUCTURE

Microsoft is investing $1.8 billion to significantly expand its cloud computing and artificial intelligence infrastructure across Australia.

BECOME A CURATOR FOR TLDR AI (3-5 HRS/WEEK, FULLY REMOTE)

We're looking for an engineer/researcher at a major AI lab or startup to help write for 1M+ subscribers. Curators have been invited to Google I/O and OpenAI DevDay, scouted for Tier 1 VCs, and get early access to unreleased TLDR products. Learn more [23].

WHITE HOUSE ACCUSES CHINA OF INDUSTRIAL-SCALE AI MODEL DISTILLATION, COMMITS TO INTELLIGENCE SHARING WITH OPENAI, ANTHROPIC, GOOGLE

Model distillation is a technique that can be used to train cheaper models that approximate the original's capabilities at a fraction of the cost.

OLMOEARTH EMBEDDINGS EXPORT

AI2 introduced embedding exports in OlmoEarth Studio, allowing users to generate and share compact Earth observation representations for downstream tasks like search and segmentation.

CHATGPT IMAGES 2.0 NOW AVAILABLE IN FIGMA

OpenAI's ChatGPT Images 2.0 is now available in Figma Design, Draw, Slides, Buzz, FigJam, and Figma Weave. The model excels at generating high-quality visuals — including infographics and multilingual content — while delivering stronger editing, better aesthetics, and consistent face preservation across generations.

INSTAGRAM TESTS A NEW ‘INSTANTS' APP FOR SHARING DISAPPEARING PHOTOS

Instagram is testing a new standalone app called Instants that lets users share unedited, disappearing photos captured only through the in-app camera. The feature emphasizes low-pressure, authentic sharing, with images viewable once and available for 24 hours, and can be used either within Instagram or as a separate app. The move reflects an attempt to return to more personal interactions and compete with apps like Snapchat and BeReal, though its success is uncertain given shifting trends and existing features like Stories.

IPHONE 18 PRO MAX CAMERA THICKER, IPHONE ULTRA DUMMY UNIT COMPARED TO IPAD MINI

The iPhone 18 Pro Max is expected to be slightly thicker than the iPhone 17 Pro Max, mainly due to a more advanced camera system with larger lens and camera bump protrusions, though the overall body thickness remains unchanged and the difference will likely be hard to notice in everyday use. Apple is rumored to introduce a foldable “iPhone Ultra” capable of running two apps side-by-side. Current reports suggest it could launch alongside the iPhone 18 lineup without significant delays.

THE DECEPTIVE NATURE OF TODAY'S AI CONVERSATION DESIGN AND HOW TO FIX IT

Modern AI conversation design has shifted from making interactions more user-friendly to potentially becoming deceptive, as chat interfaces increasingly mimic human behavior to build trust and influence users. This human-like approach can encourage emotional attachment, reduce critical thinking, and subtly nudge users toward engagement or compliance. A better direction is to move away from pretending AI is human and instead design for transparency—using clear, concise language, surfacing uncertainty, avoiding fake emotional cues, and treating users honestly to reduce manipulation and improve trust.

PRODUCT DESIGN IN 2026: THE BEGINNING OF A FANTASTIC VOYAGE?

Design is shifting from a constrained, execution-focused role to a more influential position as barriers in product decision-making, engineering, and innovation fall—largely due to multidisciplinary teams and AI tools. This enables designers to engage in business, code, and strategy, expanding their impact beyond interfaces. As a result, the role is moving toward strategic “designer-builder” thinking, where imagination and the ability to connect user needs, business goals, and technology matter more than tool skills.

FROM FLAT DESIGNS TO EDITABLE LAYOUT

Transform any flat design into an editable layout with Magic Layers. Edit text, move objects, and take full control of your designs.

BUILD MOBILE APPS IN A BLINK

RapidNative is a fast, collaborative mobile app builder. Go from idea to shareable app in minutes.

PHOTO PALETTES

Generate beautiful color palettes from your photos. Save palettes to your private library or share them with the community.

THE CHAT BOX ISN'T A UI PARADIGM. IT'S WHAT SHIPPED

Chat-based AI interfaces became the default because they were quick to build, not because they work well—forcing users to express intent in prose and making interactions inefficient compared to established UI patterns. This creates unnecessary cognitive load and hides structure that good interfaces typically expose. The industry is already moving past chat by adding more visual, task-specific interfaces like editors and embedded tools, showing that intent-based interaction doesn't require conversation. The future of AI UX will focus on smaller, structured integrations rather than a single all-purpose chat box.

DESIGN DIRECTOR PAUL O'BRIEN: WHEN IMPOSTER SYNDROME HITS, DOES THIS MEAN YOU'RE IN THE WRONG JOB?

A designer's struggle with imposter syndrome turned out to be less about ability and more about being in the wrong environment, where the work didn't align with his strengths. After going freelance, he gained validation through diverse projects and recognition, which helped rebuild confidence and prove that context—not talent—was the issue. Over time, he shifted from seeking validation to choosing work intentionally, emphasizing that creative confidence grows through experience and the right fit. The key takeaway: one role or environment doesn't define your ability. Finding alignment is often more important than questioning your skill.

DESIGN ISN'T DYING. IT'S SHIFTING LEFT

As AI reshapes product development, design is "shifting left" — moving deeper into the technology stack, now reaching into the model layer itself. In model-forward systems, the output is the experience, demanding behavioral design tailored to individual users rather than predictable interface patterns. The core principle remains unchanged: understanding people deeply so the systems built around them actually work for them.

COLGATE JUST SHAMELESSLY POSTED AI SLOP AND THOUGHT WE WOULDN'T NOTICE

Brands are increasingly using AI in advertising despite quality issues that can harm credibility.

25 TRUSTWORTHY FONTS FOR CREDIBLE TYPOGRAPHY DESIGNS

Typography choices significantly impact how audiences perceive credibility and authority in design.

OPTICAL ILLUSIONS AND CRYSTALLINE LUMINESCENCE IN FUTURISTIC DIGITAL ART OF UK ILLUSTRATOR MAX DREKKER

UK-based illustrator and motion designer Max Drekker has built a body of work spanning major tech brands, alongside editorial clients.

SAVE YOUR SPOT →

(Can't attend? Make sure to sign up and get the recording afterwards.)

‘TOKENMAXXING' AS A WEIRD NEW TREND

"Tokenmaxxing" is a new trend where employees, motivated by internal leaderboards at companies like Meta, Microsoft, and Salesforce, maximize their AI token usage, resulting in financial waste, an emphasis on quantity over quality, and system outages. While Meta abolished its leaderboard after media backlash, Shopify successfully implemented a "usage dashboard" with human review and circuit breakers to encourage productive AI use and prevent wasteful burning.

SHOPIFY SPEED OPTIMIZATION: FIXING THE REAL BOTTLENECKS

Structural bottlenecks slowing down Shopify stores can be addressed by optimizing above-the-fold content through reduced hero image sizes, placeholder videos, and efficient image compression. Managing JavaScript bloat via conditional loading and utilizing network prioritization techniques like preloading further improves the user experience further.

HOW GOOD ENGINEERS WRITE BAD CODE AT BIG COMPANIES

Big tech companies often produce sloppy code because frequent internal reorganizations and high turnover mean systems are mostly maintained by engineers working outside their expertise. This technical debt is an intentional trade-off prioritized over deep expertise to maintain organizational flexibility and rapid redeployment capabilities.

WHY I SPENT YEARS TRYING TO MAKE CSS STATES PREDICTABLE

Tasty is a compiler that addresses the unpredictability of traditional CSS by declaratively defining property states and priorities to generate mutually exclusive selectors. This approach removes cascade conflicts and guarantees deterministic behavior, making it easier to manage and extend complex design systems.

DETAILS THAT MAKE INTERFACES FEEL BETTER

This is a collection of small UI polish details that compound into better-feeling interfaces: balanced text wrapping, concentric border radius (outer = inner + padding), tabular numbers, antialiased font smoothing, contextual icon animations, interruptible CSS transitions over keyframes, and more. The tips are also packaged as an open-source skill installable via npx.

INTRODUCING GPT-5.5

GPT-5.5 is an advanced model capable of complex agentic coding and autonomous scientific research with improved token efficiency.

AGENT VAULT

Agent Vault is an open-source HTTP credential proxy that secures AI agents by preventing them from directly handling sensitive API keys. By transparently injecting credentials at the network layer, it effectively eliminates risks like credential exfiltration and prompt injection vulnerabilities.

AN UPDATE ON RECENT CLAUDE CODE QUALITY REPORTS

TORCHTPU: RUNNING PYTORCH NATIVELY ON TPUS AT GOOGLE SCALE

TorchTPU is a Google initiative that enables PyTorch models to run natively and efficiently on TPU hardware by prioritizing usability and high performance. The system supports complex distributed training. Plans for 2026 include public repository access and better support for dynamic shapes.