Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
The Curve is a conference where the accelerationists and the worried come together to talk. This year focused on where AI was on the technological Richter scale and how it will change the world. The quality of talkers this year was high, and the schedule forced hard choices between sessions. This post describes what it is like attending the conference and summarizes some of the discussions held.
Gemini Robotics 1.5 integrates AI agents into the physical world. This advancement showcases AI's growing potential in robotics by enhancing real-world interactions.
Sponsored
Everyone is scrambling to figure out AI monetization, and cost-plus credit models have become a common default. This model can work - but
OpenAI is developing a playbook for funding its eye-watering compute deals through circular financing.
monday.com built an AI-powered migration system called Morphex to break apart its massive JavaScript monolith.
xAI's financing will be split between about $7.5 billion of equity and as much as $12.5 billion of debt.
Intercontinental Exchange, parent company of the New York Stock Exchange, has invested $2 billion cash into Polymarket at a $9 billion valuation, marking the prediction market's transition from niche crypto platform to mainstream finance. The two companies will collaborate on tokenization initiatives, and ICE will distribute Polymarket's data. The investment capitalizes on growing prediction market enthusiasm as Polymarket became a political reporting staple during the 2024 election, with daily volumes peaking at $400 million, though volumes have since normalized to $40-80 million in early October as the platform and competitor Kalshi blur lines between sports prediction markets and sports betting.
Bitcoin pulled back 3% to $122,000 from record highs above $126,000 with major altcoins suffering steeper declines of 5-7% including XRP, DOGE, ADA, and AVAX, as analysts warned the crypto rally shows signs of short-term overheating following a near-vertical 16% pump off late September lows. K33 Research noted the past week marked the strongest BTC accumulation of the year with 63,083 BTC ($7.7 billion) added across US ETFs, CME, and perpetual futures, surpassing May's peak and creating elevated risk for consolidation as "similar bursts in exposure have often coincided with local tops." The selloff rippled through crypto stocks with Strategy down 7%, Coinbase down 4%, and ether treasury companies Bitmine and Sharplink falling 3-7%.
Brex will become the first global corporate card to enable instant balance payments with stablecoins, allowing business customers to accept stablecoins with automatic USD conversion, send stablecoins directly from USD balances, and pay card balances with digital assets starting with USDC in the coming months. The platform addresses the gap where stablecoins are increasingly used for cross-border business transactions but remain unavailable at most banks and institutional payment platforms, offering instant zero-fee payments that settle in seconds, 24/7 global reliability with instant cross-border settlement, and consolidated management within the existing Brex platform rather than requiring fragmented workflows across multiple platforms.
Keel launches as Solana's capital engine, powered by Sky (formerly MakerDAO), to address the blockchain's liquidity sequencing challenge, where onchain finance needs scale to attract large capital pools, but large capital requires existing scale to deploy effectively. The protocol serves as an onchain capital allocator with an initial $2.5 billion roadmap, utilizing over-collateralized debt and Sky's proven infrastructure to kickstart capital allocation flywheels and provide multiplicative force for ecosystem growth. Keel focuses on five areas: deeper stablecoin lending markets enabling larger borrowing sizes with greater capital efficiency, enhanced stablecoin swap depth and peg stability, TVL bootstrapping for new issuances and DeFi products, predictable savings yields with native staking and fee-free redemption, and RWA support through large-scale allocation into tokenized yield-bearing assets while improving onchain liquidity and redemptions.
Stablecoin issuance is easy, but adoption depends on leverage (custody, UI control, and liquidity). A wave of platforms, from wallets to exchanges, is launching their own stablecoins to capture yield from user balances, yet most will struggle to circulate beyond their own ecosystems. In open environments like Hyperliquid or Solana, liquidity favors incumbents like USDC and USDT. In closed systems like PayPal or Binance, custody gives issuers the leverage to impose adoption. The stablecoin future may be abundant, but fragmented.
EigenLayer founder Sreeram Kannan expands on Aaron Wright's viral thread, agreeing that stablecoins enable global, permissionless custody, but warns they pose a “counter-sovereignty risk” for nation-states. He predicts USD stablecoins will face regulatory choke points abroad, leading to local stablecoin dominance in domestic markets. As users recognize the geopolitical exposure of dollar-backed stables, Kannan argues they'll increasingly exit not to USDC or USDT, but to censorship-resistant assets like ETH and BTC.
The future of stablecoin adoption lies not in billion-dollar bank integrations, but in enabling small, everyday payments, like tipping a gamer, paying freelancers, or remitting $20 to family. These use cases, often uneconomical for traditional finance, are already gaining traction on low-fee chains like Polygon. To truly scale, the crypto industry must stop chasing institutional validation and build for the billions who need fast, cheap, global payments today.
A Polygon activist investor is spearheading a campaign to overhaul Polygon's POL tokenomics, citing a 46% price drop and underperformance relative to ETH and other L2 tokens. Their proposal calls for eliminating POL's 2% annual inflation and introducing buybacks or burns to counter sell pressure and restore investor confidence. The post has gained traction on social media and prompted responses from Polygon leadership as community debate intensifies around validator funding and supply strategy.
While the GENIUS Act cements stablecoins as narrow-bank instruments backed by Treasuries, it bars issuers from paying yield to users, creating strong incentives for platforms to issue their own coins and capture the yield themselves.
Layer-1 valuations are driven less by intrinsic models and more by perception, momentum, and market positioning.
By combining low-friction KYC, instant global payments, 24/7 markets, DeFi access, yield on idle funds, and self-custody into a unified 'super app', neobanks may become the default financial interface for the crypto-native economy.
Nine major platforms offer crypto-enabled cards, including Plasma One's 10% yields with 4% cashback across 150 million merchants, Tria's 6% kickbacks with auto-swapping functionality, EtherFi's 3-5% rewards plus ETHFI tokens, Revolut's Ultra membership with 1% cashback and premium benefits, Gnosis Pay's 4% cashback on over $100 million payment volume, and others from Coinbase, MetaMask, and Crypto.com offering various reward structures.
ChatGPT users can now generate FigJam diagrams—including flowcharts, gantt charts, state diagrams, and sequence diagrams—directly from conversations, documents, and photos. The feature transforms these inputs into collaborative starting points within FigJam. It is currently available for Free, Pro, and Plus plan users, with Enterprise, Business, and EU support coming soon.
Dbrand has released new skins that mimic Apple's Cosmic Orange iPhone 17 Pro color, allowing users to apply the shade to various devices, including MacBooks, iPads, AirPods, Android phones, and Windows laptops. Playfully calling them “knock-offs,” the company offers the skins for prices ranging from about $25 to $65, depending on the device.
Adidas has unveiled Trionda, the official match ball for the 2026 FIFA World Cup, which features red, blue, and green swirls that symbolize the host nations, Canada, the USA, and Mexico. The four-panel design includes embossed national motifs, enhanced surface texture for stability and grip, and a new side-mounted motion sensor chip that provides real-time ball data for VAR decisions.
LLMs excel at understanding intent and manner, making them powerful tools for creating adaptive interfaces that respond to user requests with appropriate UI components rather than text. This article demonstrates how to build "call-and-response UI" systems where the LLM acts as a design decision-maker, interpreting user intent and returning structured data that tells the interface which component to display, essentially translating natural language into machine-readable interface instructions. By constraining outputs to predefined UI patterns and utilizing prompts as collaborative design specifications, teams can create experiences that adapt to user needs while leveraging LLMs' strengths in understanding context and mitigating their weaknesses in handling facts.
Generative AI is transforming UX design by automating execution tasks, such as creating prototypes and analyzing feedback, allowing designers to shift from pixel-pushing to strategic roles focused on systems thinking, ethical oversight, and prompt engineering. The technology serves as a collaborative tool across all design phases, but it requires human judgment to ensure that outputs are inclusive, contextually appropriate, and aligned with user needs. UX designers must develop AI literacy and master prompt crafting while maintaining responsibility for critical thinking, as their irreplaceable value lies in asking the right questions rather than merely producing artifacts quickly.
Creatives today need both artistic skill and business savvy to thrive. The key to success is pricing work confidently, making payments effortless, diversifying sales channels, building credibility through curated portfolios and testimonials, and keeping admin simple to reduce stress, allowing more time for creativity and growth.
Run parallel Claudes safely in containers. Jump between their environments to instantly test changes. Get suggestions that catch critical issues as you go.
Pics.io is an AI-powered digital asset management platform that lets you centralize, manage, and distribute all your digital content in one place with
Create soft UI neumorphism effects with visual controls. Perfect for modern, minimal interface designs.
Traditional navigation is dying as interfaces shift toward intent-based design, where systems anticipate user needs rather than requiring explicit menu choices—seen in platforms like Spotify, TikTok, and ChatGPT. While this approach excels at routine tasks, it risks stripping users of cognitive maps and autonomy, creating disorienting "black boxes" that trap people in algorithmic bubbles and fail when predictions are wrong. The solution lies in predictive systems that augment, rather than replace, traditional navigation, offering intelligent suggestions while maintaining a transparent structure and user control.
Figma's 2025 AI report reveals that AI is fundamentally reshaping design workflows by improving speed and efficiency, though 60% of designers don't yet see quality improvements. The research shows that design skills are becoming increasingly valuable for AI products, with most AI builders considering design at least as critical as for traditional products. However, the unpredictability of AI introduces new complexities in prototyping, user testing, and determining transparency levels. Designers currently use AI primarily for research tasks, such as desk research (38%) and data analysis (40%), rather than for creative work.
Absolut Vodka has launched a new artist collaboration celebrating Pop Art legend Keith Haring, transforming London's Charing Cross station into “Haring Cross” — an immersive public art gallery featuring Haring's original Absolut works. The Playful Underground campaign continues both Haring's and Absolut's mission to make art accessible to everyone, blending brand heritage, authenticity, and the spirit of Pop Art with an emphasis on joy and inclusivity.
Companies covered by the European Accessibility Act must publish an accessibility statement explaining how their services meet the requirements and train their support staff to answer accessibility-related questions.
Creative freelancers juggling multiple income streams commonly make five critical mistakes when handling taxes, including failing to track every payment, regardless of size, and not understanding HMRC's requirement to declare all income on Self Assessment returns, even with PAYE employment.
Designers each have a personal object they can't work without—whether it's a chair, notebook, mug, pencil, or book—reflecting how creativity is shaped by comfort, habit, and personal connection as much as by tools or technology.
Sponsored
Generative AI applications require speed, scale, and reliability at low cost. Redis Cloud on AWS [4] unifies client-side caching with your primary data store to deliver high-throughput applications directly from your AWS environment.
The 2025 software engineering job market is experiencing a paradox: companies are receiving a flood of applications (often 1,000+ per role) but are struggling to find quality candidates, with most actual hires coming from sourcing and referrals rather than inbound applications. The market shows increased demand for product engineers and founding engineers, with remote positions facing more competition but offering slightly lower compensation. Also, issues like AI-generated fake applications and interview cheating are becoming growing problems.
Microlandia is a city builder game inspired by SimCity with a focus on realistic simulation using real-world data. It was initially built in Go, but the project faced challenges in creating a 3D view and a full-featured UI. To overcome these hurdles, the game is being ported to Deno, taking advantage of its cross-platform capabilities and integration with web technologies like ThreeJS and React.
A developer optimized Node.js performance for processing a 14.8GB weather data file containing 1 billion temperature measurements to achieve a 78% speedup. The main breakthrough involved bypassing JavaScript's string encoding overhead by parsing raw bytes directly, storing temperatures as integers (e.g., 25.3°C as 253 tenths), and implementing a hash-first approach that only decoded station names when absolutely necessary.
This dev uses OpenAI's GPT-5-Codex to automate AI research, and he found that Codex outperformed his solo efforts, especially in generating coherent stories. The process involved a feedback loop where Codex suggested experiments based on results, and Goedecke selected or contributed ideas. The most successful approach was distilling a transformer model from an n-gram model, which improved story coherence by initially focusing on grammatical correctness.
BUILD files in Bazel are more than just dependency lists and should be used to encode a project's high-level architecture. By manually managing BUILD files and defining larger conceptual targets, developers can enforce module boundaries, detect architectural violations early, and guide AI tools better.
Slack implemented a "Deploy Safety Program" in 2023 to address the fact that 73% of customer-facing incidents were caused by their own code deployments, as the platform became increasingly mission-critical for users. The program focused on automated detection and remediation within 10 minutes. By January 2025, the initiative successfully reduced customer impact hours by 90% from peak levels through a combination of automated rollback systems, improved monitoring, and cultural changes across their engineering organization.
Google DeepMind has introduced the Gemini 2.5 Computer Use model, which allows agents to interact directly with user interfaces. This model, available via the Gemini API, can navigate web pages and applications by clicking, typing, and scrolling, outperforming existing solutions with lower latency. The model analyzes user requests and screenshots to generate UI actions, operating within a loop for iterative task completion while prioritizing safety through built-in features and developer controls.
Arc is a high-performance time-series data warehouse built on DuckDB, Parquet, and MinIO. It supports high-speed data ingestion and fast analytical queries with SQL.
/dev/push is an open-source platform that offers a self-hostable alternative to services like Vercel. It supports Git-based deployments, environment management, team collaboration, custom domains, and real-time monitoring.
Successful AI agents in production depend on good context engineering, governance, memory architecture, and multi-model orchestration. Most AI deployments fail due to bad scaffolding around the models. Also, trust in AI is built through human-in-the-loop design and consistent, explainable behavior, positioning AI as an assistant rather than an autonomous decision maker.
Vibe Engineering is a responsible and stronger approach to software development using AI, whereas "vibe coding" is more irresponsible. Vibe engineering has experienced engineers using AI tools like coding agents to speed up their work while still having high standards. This approach requires strong software engineering practices such as automated testing, planning, documentation, and code review, while also relying on QA, research, and careful management of AI assistants.
An AI developer accidentally deleted working code but was able to recover it using an LLM with a 1 million token context window, showing an unexpected benefit of LLMs remembering past interactions.
Pdoc is a tool that automatically generates API documentation for Python projects with features like type annotation support, cross-linking, a built-in web server, and compatibility with various docstring styles.
Email is a crucial digital asset, and you can gain control over it and become "unbannable" by using your own domain with a trusted email provider and backing up all emails to a local server.
This is a mathematical guide to gradient descent during deep learning, which is an optimization algorithm used to train machine learning models by finding the minimum of a cost function.