Search the archive

WHY DON'T WE GET MORE SCIENTIFIC BREAKTHROUGHS?

Real progress is messy, iterative, and deeply intertwined with the tools we build.

APPLE DEPARTURES POINT TO CHALLENGES FOR IPHONE'S DOMINANCE

A long-term brain drain has robbed Apple of innovators and seeded rivals with expertise.

APPLE'S ATTEMPT AT SYSTEM-WIDE FILTERING API — IS IT GOOD? ADGUARD'S RESEARCH

System-wide filtering is the ability to intercept and block web traffic from any application, not just browsers.

HOW FOSS WON AND WHY IT MATTERS

Free and open source software is an amazing hack for cost control and copyright licensing law problems.

APPLE ANNOUNCES WINNERS OF THE 2025 APP STORE AWARDS

The iPhone app of the year was Tiimo, the iPad app of the year was Detail, and the Mac app of the year was Essayist.

THE RAM SHORTAGE COMES FOR US ALL

AI datacenter buildouts are causing RAM prices to skyrocket.

Download the free guide (no form fill required)

ANTHROPIC INTERVIEWER

Anthropic Interviewer is a tool that uses AI to conduct and analyze large-scale interviews to research AI's role in work across different professions. Initial findings from 1,250 professionals showed optimism toward AI enhancing productivity while highlighting concerns over job displacement and security in creative and scientific fields. Anthropic plans to use this data to improve AI models and influence policy, collaborating with artists, scientists, and educators to align AI development with user needs.

GOOGLE PARTNERS WITH REPLIT, IN VIBE-CODING PUSH

Google Cloud has partnered with AI coding startup Replit to enhance enterprise vibe-coding using Google models. Replit will use Google Cloud services to expand its platform, supporting AI coding for enterprise clients. This collaboration aims to drive Google Cloud adoption and extend AI reach beyond traditional engineers.

GEMINI 3 DEEP THINK IS NOW AVAILABLE IN THE GEMINI APP

Gemini 3 Deep Think uses parallel reasoning to explore multiple hypotheses simultaneously, building on the Gemini 2.5 Deep Think variants that won a gold medal on the International Mathematical Olympiad.

STATE OF AI

This year was a turning point in the real-world use of large language models. The field shifted from single-pass pattern generation to multi-step deliberation inference. The shift up folded so fast that our understanding of how these models have been used in practice has lagged behind. This study leverages the OpenRouter platform to analyze over 100 trillion tokens of real-world AI interactions to see how the technology is being used in the real world. The way developers and end-users have been engaging with AI is complex and multifaceted. The study shows how a data-driven understanding of usage can inform better design and deployment.

WE GOT CLAUDE TO FINE-TUNE AN OPEN SOURCE LLM

Hugging Face Skills gives Claude the ability to fine-tune language models. It can submit jobs to cloud GPUs, monitor progress, and push finished models to the Hugging Face Hub. This tutorial teaches readers how it works and how to use it. The tool allows users to train models from 0.5B to 70B parameters, convert them to GGUF for local deployment, and run multi-stage pipelines that combine different techniques.

GPT-5.1-CODEX-MAX PROMPTING

OpenAI has outlined how to get optimal results from GPT-5.1-Codex-Max, highlighting its faster token efficiency, long-running autonomy, and improved compaction for extended reasoning.

→ Watch Replay

BUILDING WITH CURSOR (PUBLIC)

This is the public-facing version of an internal onboarding guide at Cursor. It walks through how to get started from scratch to a built-out and deployed project. It covers how to set up and use Cursor, how to build and customize projects, and how to deploy using Vercel.

CONTEXT ENGINEERING FOR AI AGENTS

The field of context engineering is moving fast. However, the biggest performance gains nowadays come from removing complexity. As models get stronger, we should be getting out of the model's way. Context engineering is about minding the minimal effective context required for the next step, not about adding more context.

ARCHITECTING EFFICIENT CONTEXT-AWARE MULTI-AGENT FRAMEWORK FOR PRODUCTION

The landscape of AI agent development is shifting fast. Organizations are now deploying sophisticated, autonomous agents to handle long-horizon tasks. However, this ambition is being bottlenecked by context. The context stack in Google Agent Development Kit was developed to support context engineering. The open-source, multi-agent-native framework is built to make active context engineering achievable in real systems.

THE MATH LEGEND WHO JUST LEFT ACADEMIA—FOR AN AI STARTUP RUN BY A 24-YEAR-OLD

Ken Ono is one of the most prominent mathematicians in the world. He recently joined Axiom Math to revolutionize math with AI. The company was founded by one of his former students. He joined because he couldn't resist the opportunity to put his mark on something other than a chalkboard.

POWER OVERWHELMING

AI capex is driving US GDP growth, yet a $1.5 trillion AI revenue shortfall looms compared to invested capital. OpenAI's infrastructure spend and emerging AI application revenues like ChatGPT's $20B by 2025 reveal a disconnect between projected and necessary earnings to justify current investments. The uncertain AI cloud business model, characterized by rapid hardware obsolescence, suggests heavy reliance on the Magnificent Seven's internal workloads to prevent an impending AI bubble and excessive market overbuild.

NVIDIA AND AWS EXPAND FULL-STACK PARTNERSHIP, PROVIDING THE SECURE, HIGH-PERFORMANCE COMPUTE PLATFORM VITAL FOR FUTURE INNOVATION

NVIDIA and AWS expanded their partnership to integrate NVIDIA NVLink Fusion with AWS' Trainium4, Graviton CPUs, and Nitro System, enhancing cloud-scale AI capabilities.

FOR AI AT WORK, START WITH SOMETHING SMALL

AI tools excel at automating repetitive, time-consuming tasks in the "messy middle," such as summarizing notes or categorizing feedback.

ANTHROPIC INTERVIEWER (HUGGING FACE REPO)

Anthropic Interviewer is a tool for conducting AI-powered qualitative research interviews at scale.

AI #145: YOU'VE GOT SOUL

The cycle of language model releases is now complete.

THE FIRST RESEARCH ARTICLE IN THEORETICAL PHYSICS IN WHICH THE MAIN IDEA CAME FROM AN AI

The paper derives new operator integrability conditions required for foliation independence.

CRYPTO MARKETS HOLD GAINS AS BITCOIN TRADES ABOVE $92K

Bitcoin stabilized above $92,000 as Ethereum led top-20 gains at 3.4%. $732 billion in new capital was added this cycle. 1-year realized volatility nearly halved, suggesting a "calmer, larger and more institutional" environment. Over 25% of BTC supply remains underwater as spot Bitcoin ETFs recorded $14.9 million outflows, breaking a five-day streak. Liquidations totaled $282 million, with shorts accounting for $201 million as traders position ahead of the Fed's expected December 10 rate cut and digest jobless claims falling to 191,000.

MONEYGRAM TEAMS UP WITH FIREBLOCKS TO EXPAND STABLECOIN PAYMENTS

MoneyGram partnered with Fireblocks to power stablecoin payments and cross-border settlements across its 200+ country network, reducing pre-funding requirements and enabling real-time treasury operations through multi-blockchain stablecoin transfers. The integration allows near-instant digital wallet arrivals for recipients using USDC while MoneyGram reconciles payments faster on the backend, reducing friction from local banking systems and capital requirements. Fireblocks, which has secured over $5 trillion in annual digital asset transfers, will act as the programmable layer routing value across chains and jurisdictions.

ETHVAL V0.4.3: TRACKING FUSAKA'S SUPPLY EXPERIMENT

ETHval just shipped a “Supply Dynamics” dashboard to watch what Fusaka does to ETH inflation/deflation in real time. Fusaka boosts Ethereum's gas limit from 36M to 60M and massively expands blob capacity, so fees per transaction should fall at first. It is still debated whether cheaper blockspace makes ETH inflationary (because burn drops), or whether it pulls in so much extra activity that total burn still beats issuance, and ETH stays scarce. ETHval now surfaces gas utilization, daily burn/issuance, net supply, base fee, blob activity, and TVL so you can see which way the experiment goes.

BASE LAUNCHES BRIDGE TO SOLANA

Base launched a bridge to Solana, powered by Chainlink's Cross-Chain Interoperability Protocol, enabling native SPL token support in Base apps and bidirectional asset transfers verified by Coinbase and Chainlink node operators. The integration advances Base's goal to serve as a hub for the "everything economy" where users can trade and use assets across chains regardless of their starting network.

TETHER IN 2025: A CAPITAL ANALYSIS

Tether is like an unregulated bank rather than a simple money transmitter. Depending mostly on how harshly you treat BTC, Tether's total capital ratio ranges from ~10.9% (OK vs minimal regulatory standards) down to ~3.9% (bad vs market norms). Group-level profits and equity look big, but most sit outside the actual reserve entity and aren't legally committed to back USDT redemptions, so they can't be counted as true regulatory capital.

WHAT ACTUALLY HAPPENS WHEN YOU SUBMIT A TRANSACTION

Ethereum transaction processing moved from validators building their own blocks to specialized block builders dominating the process through MEV-Boost relays that auction optimally-ordered blocks to validators selecting the most profitable bids. Transactions propagate through decentralized mempools where each node maintains local versions and compete on fees rather than arrival time, with wallets signing and broadcasting via RPC nodes like Alchemy before builders bundle transactions to capture MEV opportunities. Confirmation arrives after one 12-second slot when the majority attests the block, while finalization requires 12.8 minutes, providing cryptographic permanence that requires one-third of staked ETH to reverse.

PEERDAS: HOW FUSAKA MAKES L2S CHEAPER

Fusaka's main feature, PeerDAS, changes how Ethereum stores blob data. Instead of every node keeping full copies of each blob, nodes only store a small slice of every blob while still being able to verify that the full data exists. That means the network can safely handle many more blobs per block, which lowers data-availability costs for rollups. More blobs per block means cheaper L2 data and cheaper L2 transactions for users.

CAN L1S COMPETE WITH BTC AS MONEY?

Most of crypto's value is really a bet on “cryptomoney”, BTC plus L1 tokens that the market hopes can act like money one day. When you strip out TRON/Hyperliquid, L1 revenues have fallen, while their price-to-sales multiples have exploded (40x to 500x+), meaning valuations are being driven far more by monetary premium than by cash flows. Against that standard, almost all major L1s have underperformed BTC since late 2022. Even Solana, whose fundamentals went from 20 to 30x, only managed an 87% outperformance (at one point) versus BTC. BTC's claim as cryptomoney is strengthening while the “maybe we'll be money later” story for most L1s is quietly breaking.

ARBITRUM'S BRIEF POST-FUSAKA FEE SPIKE SHOWS WHY CLIENTS MATTER

Right after Ethereum's Fusaka upgrade, median fees on Arbitrum One briefly spiked to about $0.99 before dropping back to the usual ~$0.005.

FUSAKA BLOB FEES & ETH BURN

Fusaka is now live, and EIP-7918 changes how “blob” data fees work on Ethereum.

EF PROTOCOL TEAM'S PRIORITIES

The Ethereum Foundation protocol team shipped production-ready Open Intents Framework, EIL trustless crosschain infrastructure, interoperable chain-specific addresses, and Fast Confirmation Rule, reducing confirmation times to one slot while targeting sub-10-second economic finality and optimistic rollup withdrawal windows shortened from seven days to 1-2 days, then one hour.

BINANCE JUNIOR

Binance has released a new account type for kids, allowing parents to set up specific permissions and monitoring while giving their children access to the platform's features.

Apply to claim your free year →

GOOGLE LAUNCHES WORKSPACE STUDIO TO CREATE AUTOMATED GEMINI-POWERED AGENTS

Google's Workspace Studio is a no-code automation tool that enables users to create AI agents powered by Gemini 3 to handle tasks across Gmail, Chat, Drive, and other productivity apps. The agents use natural-language prompts and can perform sentiment analysis, content generation, and intelligent prioritization by reasoning through problems rather than rigid rules. Workspace Studio is rolling out to Business, Enterprise, and Education tier customers over the coming weeks.

APPLE DESIGN BOSS ALAN DYE DEPARTING FOR META

Alan Dye, Apple's Human Interface Design chief since 2015, is leaving for Meta to lead a new design studio focused on hardware, software, and AI. Apple veteran Stephen Lemay will take over his role. While some commentators argue Dye's departure won't meaningfully affect either Apple or Meta, it comes during a period of major leadership turnover at Apple, including recent exits by key executives and ongoing shifts as the Tim Cook era winds down.

IOS 26.2 WILL ADD THREE NEW WAYS TO CUSTOMIZE YOUR IPHONE

iOS 26.2 introduces enhanced customization and accessibility features for iPhone users. The update expands the Liquid Glass clock slider on the Lock Screen for greater transparency control, adds a new “Screen” flash option for notifications alongside the camera flash, and brings an “Urgent” toggle for Reminders with customizable alarm behavior, letting users snooze or immediately complete tasks.

DIGITAL TWIN MODELING: A VISION OF AN AI FUTURE WITH UX AT THE HELM

Digital twin modeling—systems that create real-time behavioral models to predict and adapt to user actions—offers UX designers a concrete framework for working with AI, addressing the current confusion around "UX in AI." Unlike vague interpretations of AI design work, digital twins enable continuous learning from user behavior patterns. Learning digital twin modeling requires mastering data-informed design fundamentals first, then progressively building systems that identify behavioral patterns, automatically adapt experiences, and eventually predict user needs before problems arise.

IN THE SEA OF SAMENESS

Modern digital design has become increasingly homogeneous due to frameworks like Tailwind and design system templates, which prioritize efficiency over creative authenticity. The shift from craft-focused design to metrics-driven product development, combined with formulaic portfolio structures that reduce design to "problem-solving," has diminished the designer's role and unique voice. Designers need to reconnect with personal taste and intuition, creating space for authentic self-expression that goes beyond imitation and trend-following.

THE ONLY WINNING MOVE IS NOT TO PLAY

User research relies on human expertise, judgment, and context-gathering skills that AI cannot replicate. While AI can assist with tasks like pattern recognition or transcription, outsourcing entire research plans, interviews, or synthesis risks producing average results, devaluing the profession, and misleading organizations into thinking they are truly user-focused. Maintaining professional standards and preserving the uniquely human aspects of research is essential, resisting the temptation to trade expertise for efficiency or hype-driven AI adoption.

THE EASIEST WAY TO BUILD A WEBSITE

Design, edit, and publish a beautiful website with absolutely no code.

PROMPT YOUR NEXT APP

Turn ideas into live apps fast with MeDo—AI-driven, no coding, ready-to-use plugins.

GEMINI 3 BRAND AUDIT

Find out what Gemini 3 thinks about your brand and competitors.