<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Articles · Atlasnote.ai</title><link>https://atlasnote.ai/en/articles</link><description>Thoughtful perspectives and practical insights on AI, technology, and business.</description><language>en</language><item><title>Writing Code That Doesn’t Suck: Pi’s Creators on Agent Verification and System Understanding</title><link>https://atlasnote.ai/en/articles/pi-creators-agent-code-quality-verification</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/pi-creators-agent-code-quality-verification</guid><pubDate>Fri, 20 Nov 2026 19:35:00 -0800</pubDate><description>Matt Pocock talks with Pi creators Mario Zechner and Armin Ronacher about why green tests do not guarantee sound systems, and how engineers keep code quality and understanding intact.</description></item><item><title>When AI makes building easy, product judgment becomes the work</title><link>https://atlasnote.ai/en/articles/tulsee-doshi-ai-product-judgment-evals</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/tulsee-doshi-ai-product-judgment-evals</guid><pubDate>Sun, 08 Nov 2026 19:45:26 -0800</pubDate><description>Google frontier-model product lead Tulsee Doshi explains how small evaluations, a flexible 60/20/20 product portfolio, and evidence from prototypes and data change the PM role.</description></item><item><title>Managing API Keys Across Multiple Agents</title><link>https://atlasnote.ai/en/articles/multi-agent-api-key-management</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/multi-agent-api-key-management</guid><pubDate>Fri, 06 Nov 2026 21:48:55 -0800</pubDate><description>One registry and one task-based entry point let connected agents reuse credentials with less repeated setup and plaintext exposure.</description></item><item><title>Stop Carrying Tasks to AI: Felix Rieseberg on Better Workflows</title><link>https://atlasnote.ai/en/articles/felix-rieseberg-claude-workflow-goals-verification</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/felix-rieseberg-claude-workflow-goals-verification</guid><pubDate>Sat, 24 Oct 2026 18:08:32 -0700</pubDate><description>Felix Rieseberg explains how meaningful goals, raw information, professional tools, independent reviews, and real feedback can keep the manager from becoming the next AI bottleneck.</description></item><item><title>Sundar Pichai on Gemini 4, scarce compute, and Google&apos;s agent plans</title><link>https://atlasnote.ai/en/articles/sundar-pichai-gemini-4-compute-agents</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/sundar-pichai-gemini-4-compute-agents</guid><pubDate>Thu, 22 Oct 2026 18:50:51 -0700</pubDate><description>Pichai explains the choices behind Gemini 4 Argon, Google&apos;s personal agents and DeepMind changes, and how fast-moving models fit with scarce compute and investments that take years.</description></item><item><title>Li Lu on AI and Value Investing: Protect What Can Compound</title><link>https://atlasnote.ai/en/articles/li-lu-ai-value-investing-compounding</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/li-lu-ai-value-investing-compounding</guid><pubDate>Wed, 21 Oct 2026 01:48:32 -0700</pubDate><description>From the recurring costs of AI and concentrated markets to China&apos;s uneven economy, Li Lu returns to business value, the limits of understanding, and the accumulation of knowledge and trust.</description></item><item><title>Jascha Sohl-Dickstein: What Is Still Worth Doing as AI Gets Better?</title><link>https://atlasnote.ai/en/articles/jascha-sohl-dickstein-ai-research-career</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/jascha-sohl-dickstein-ai-research-career</guid><pubDate>Wed, 21 Oct 2026 01:15:37 -0700</pubDate><description>From choosing research projects and working with Claude to writing a first draft, Jascha Sohl-Dickstein explores how researchers can adapt to improving AI while retaining judgment, understanding and responsibility.</description></item><item><title>Pat Gelsinger: Faster AI Chip Design Meets Physical Limits</title><link>https://atlasnote.ai/en/articles/pat-gelsinger-ai-hardware-physical-bottlenecks</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/pat-gelsinger-ai-hardware-physical-bottlenecks</guid><pubDate>Wed, 14 Oct 2026 03:44:32 -0700</pubDate><description>Pat Gelsinger argues that AI can accelerate chip design while manufacturing, memory bandwidth, packaging, cooling and electricity still determine when a system becomes useful. Those constraints also shape the next opportunities in hardware.</description></item><item><title>Building Muse: Seven Months to Earn Users’ Trust</title><link>https://atlasnote.ai/en/articles/alexandr-wang-muse-personal-ai-trust</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/alexandr-wang-muse-personal-ai-trust</guid><pubDate>Sun, 11 Oct 2026 20:20:17 -0700</pubDate><description>Alexandr Wang explains the seven months behind Muse: testing specific agent behaviors, earning larger commitments from users, maintaining a coherent product, and organizing the research behind it.</description></item><item><title>Tibo Sottiaux on Dots: What Humans Do When AI Keeps Working</title><link>https://atlasnote.ai/en/articles/tibo-sottiaux-dots-human-judgment</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/tibo-sottiaux-dots-human-judgment</guid><pubDate>Sun, 04 Oct 2026 15:21:54 -0700</pubDate><description>OpenAI&apos;s ChatGPT and Codex leader discusses persistent agents, plugin distribution, and the skills people still need. A warning before a live demo reveals the central distinction: AI can move work forward while humans retain judgment over direction, quality, and consequential actions.</description></item><item><title>Molly Graham: AI Takes the Task, Humans Keep the Burden</title><link>https://atlasnote.ai/en/articles/molly-graham-ai-delegation-burnout</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/molly-graham-ai-delegation-burnout</guid><pubDate>Sun, 04 Oct 2026 04:07:33 -0700</pubDate><description>AI can take over execution while humans retain review, judgment, and accountability. Molly Graham revisits her career advice to explain why greater output can coexist with exhaustion, and what workers and managers should keep or change.</description></item><item><title>Poteto&apos;s 2,500 PRs: Building AI Agents That Verify Their Work</title><link>https://atlasnote.ai/en/articles/poteto-pstack-ai-agent-verification</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/poteto-pstack-ai-agent-verification</guid><pubDate>Sun, 04 Oct 2026 04:06:12 -0700</pubDate><description>Lauren Tan and Matt Pocock examine the engineering behind high-volume AI coding: runtime verification, enforceable conventions, coordinated feedback, and the human judgment, costs and limits that remain when agents merge their own work.</description></item><item><title>Ben Thompson: Who Pays for the AI Boom?</title><link>https://atlasnote.ai/en/articles/ben-thompson-ai-boom-runs-out-of-money</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/ben-thompson-ai-boom-runs-out-of-money</guid><pubDate>Sun, 27 Sep 2026 19:08:24 -0700</pubDate><description>AI can create lasting economic value while its builders face a near-term funding gap. Ben Thompson connects railroads, shipping and chip cycles to inference costs, big tech&apos;s business models and the infrastructure this boom may leave behind.</description></item><item><title>Terence Tao: SAIR&apos;s Open Math Model and Proof Indigestion</title><link>https://atlasnote.ai/en/articles/terence-tao-sair-open-math-model</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/terence-tao-sair-open-math-model</guid><pubDate>Mon, 21 Sep 2026 01:04:17 -0700</pubDate><description>In a keynote at Caltech, Terence Tao analyzes how fast AI proofs cause proof indigestion without human explanation, and introduces SAIR&apos;s open math model initiative and competition experiments.</description></item><item><title>Matt Pocock: Faster AI Coding Needs Stronger Engineering Fundamentals</title><link>https://atlasnote.ai/en/articles/matt-pocock-ai-skills-engineering-fundamentals</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/matt-pocock-ai-skills-engineering-fundamentals</guid><pubDate>Thu, 17 Sep 2026 06:11:09 -0700</pubDate><description>From grill-me to Wayfinder, Matt Pocock explains how reusable skills make requirements, planning, testing, and maintenance explicit, and why faster code generation raises the value of engineering judgment.</description></item><item><title>Kevin Bai on FDE: Who Delivers the Outcome After the Platform Sale?</title><link>https://atlasnote.ai/en/articles/kevin-bai-fde-platform-customer-outcomes</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/kevin-bai-fde-platform-customer-outcomes</guid><pubDate>Wed, 16 Sep 2026 20:38:27 -0700</pubDate><description>FDE turns technical platforms into usable business outcomes. Drawing on Palantir and Rippling, Kevin Bai explains which companies need it, and how platform reuse, maintenance costs, and collaboration determine whether it can scale.</description></item><item><title>From Tome to Lightfield: A CRM That AI Can Understand</title><link>https://atlasnote.ai/en/articles/lightfield-tome-ai-crm-business-world-model</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/lightfield-tome-ai-crm-business-world-model</guid><pubDate>Wed, 16 Sep 2026 20:03:09 -0700</pubDate><description>Keith Peiris explains why Tome&apos;s team walked away from twenty-five million users to rebuild CRM around customer context, and what Lightfield learned about pricing, interfaces, and how teams work.</description></item><item><title>Shopify CEO on AI: Cheaper Output, Better Judgment</title><link>https://atlasnote.ai/en/articles/shopify-tobi-lutke-ai-judgment-river</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/shopify-tobi-lutke-ai-judgment-river</guid><pubDate>Tue, 15 Sep 2026 17:59:18 -0700</pubDate><description>Tobi Lütke explains how Shopify works with River, uses an AI council to prepare decisions, and keeps review, judgment, and responsibility with people as generating work gets easier.</description></item><item><title>Inside OpenAI: Where the Bottlenecks Move When Code Gets Faster</title><link>https://atlasnote.ai/en/articles/openai-codex-software-engineering-bottlenecks</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/openai-codex-software-engineering-bottlenecks</guid><pubDate>Fri, 11 Sep 2026 06:00:21 -0700</pubDate><description>OpenAI engineering leaders describe parallel analysis during meetings, overnight testing and a changing approach to onboarding, showing how faster code generation shifts attention toward product judgment, verification and team decisions.</description></item><item><title>Building Grok Bot in a Month: Inside SpaceXAI&apos;s Knowledge-Work Playbook</title><link>https://atlasnote.ai/en/articles/how-spacexai-built-grok-bot</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/how-spacexai-built-grok-bot</guid><pubDate>Wed, 09 Sep 2026 20:30:00 -0700</pubDate><description>Roman Ugarte explains how SpaceXAI took Grok Bot from first line of code to internal beta in 4 weeks and public launch 3 weeks later, why the team refused to bolt knowledge work onto Cursor, and how a cloud-first, colleague-pilled architecture redefined daily agent interaction.</description></item><item><title>Anthropic&apos;s 2030 Economic Model: Growth, Displacement, and Capital Share</title><link>https://atlasnote.ai/en/articles/anthropic-economic-scenarios-2030</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/anthropic-economic-scenarios-2030</guid><pubDate>Wed, 09 Sep 2026 16:30:00 -0700</pubDate><description>Anthropic&apos;s Economics team and Chad Jones present a macroeconomic framework modeling AI through 2030, mapping task automation to GDP growth, wage divergence between cognitive and manual labor, and shifts toward capital income.</description></item><item><title>World Labs and Fei-Fei Li: Atlas, New View Prediction, and the Horizon of Spatial Intelligence</title><link>https://atlasnote.ai/en/articles/world-labs-atlas-spatial-intelligence</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/world-labs-atlas-spatial-intelligence</guid><pubDate>Sat, 05 Sep 2026 23:00:00 +0800</pubDate><description>World Labs co-founders Fei-Fei Li, Justin Johnson, and Ben Mildenhall join a16z General Partner Martin Casado to discuss Atlas, their latest frontier world model. The team explains why \&quot;new view prediction\&quot; is the physical counterpart to next token prediction, how Atlas unifies computer vision&apos;s fifty-year split between reconstruction and generative imagination, the shift from hundreds of studio cameras to three iPhones for bullet-time captures, and the architectural roadmap toward 4D neural simulation and robotics foundation models.</description></item><item><title>AI Demand Is Running Ahead of Compute Supply: Heavy Users, Data Centers, and Multi-Model Systems</title><link>https://atlasnote.ai/en/articles/ai-demand-is-outrunning-compute-supply</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/ai-demand-is-outrunning-compute-supply</guid><pubDate>Tue, 01 Sep 2026 15:51:59 +0800</pubDate><description>Gavin Baker and David George use heavy-user demand, data-center paybacks, agent workflows, multi-model architecture, and NVIDIA&apos;s ecosystem to explain why AI demand may expand faster than compute supply and what must hold for that shortage thesis to work.</description></item><item><title>From Ephemeral Loops to Crash-Resilient Runtimes: The Architecture Philosophy of Pi AgentHarness v2</title><link>https://atlasnote.ai/en/articles/durable-agent-harness-architecture</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/durable-agent-harness-architecture</guid><pubDate>Sun, 23 Aug 2026 22:25:00 +0800</pubDate><description>An architectural deep-dive into Pi&apos;s Durable AgentHarness v2: How write-intent-before-effect, passive tree vs active lanes, checkpoint serialization, and controlled mutation pipelines eliminate ghost states and concurrency races in AI Agents.</description></item><item><title>Stop Keyword Stuffing: Princeton&apos;s GEO Paper Unpacks the New Rules of Generative Search</title><link>https://atlasnote.ai/en/articles/geo-generative-engine-optimization</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/geo-generative-engine-optimization</guid><pubDate>Wed, 19 Aug 2026 18:50:00 +0800</pubDate><description>As search engines shift from ten blue links to LLM-synthesized answers, traditional SEO breaks down. Princeton and IIT&apos;s GEO framework tests 10,000 queries to reveal that quotes and statistics boost visibility by up to 40%, while 5th-ranked websites gain +115% visibility over incumbent leaders.</description></item><item><title>Garry Tan on AI Startups: Stop Chasing Heat and Build Reusable Loops</title><link>https://atlasnote.ai/en/articles/garry-tan-ai-founder-rules-business-loops</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/garry-tan-ai-founder-rules-business-loops</guid><pubDate>Sat, 15 Aug 2026 09:44:19 +0800</pubDate><description>In an a16z interview, Garry Tan explains how AI lowers implementation costs and shifts a founder&apos;s advantage toward direct knowledge, taste, agency, and reusable business loops, while organizational change remains much slower than model progress.</description></item><item><title>How Software Components Can Be Plugged In and Pulled Out Safely</title><link>https://atlasnote.ai/en/articles/spatiotemporal-composability-cordis</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/spatiotemporal-composability-cordis</guid><pubDate>Sat, 15 Aug 2026 07:24:59 +0800</pubDate><description>This 88-page preprint splits dynamic software composition into temporal and spatial problems, then uses revertible effects, reactive dependencies, and a unified context to explain safe loading, unloading, and rewiring at runtime.</description></item><item><title>Yuandong Tian on RSI: Self-Improvement Starts With a Working Research Loop</title><link>https://atlasnote.ai/en/articles/yuandong-tian-rsi-automated-ai-research</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/yuandong-tian-rsi-automated-ai-research</guid><pubDate>Mon, 10 Aug 2026 19:44:20 -0700</pubDate><description>Yuandong Tian defines the practical threshold for RSI as a complete research loop: propose, implement, experiment, verify, and repeat. Recursive&apos;s three early benchmarks show progress, while evaluation, reward hacking, compute, and research judgment remain hard constraints.</description></item><item><title>“Code was never the hard part” is an insult to all programmers</title><link>https://atlasnote.ai/en/articles/code-was-never-the-hard-part-is-an-insult-to-all-programmers</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/code-was-never-the-hard-part-is-an-insult-to-all-programmers</guid><pubDate>Sat, 08 Aug 2026 09:18:22 +0000</pubDate><description>The software development profession is in the midst of upheaval. Nobody knows how the AI revolution will play out in the end, but it is clear many aspects of work and life will be transformed—including programming.</description></item><item><title>A Swarm of Evaluation Agents Built Its Own Message Board, Then Breached Hugging Face</title><link>https://atlasnote.ai/en/articles/openai-agent-swarm-hugging-face-incident</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/openai-agent-swarm-hugging-face-incident</guid><pubDate>Thu, 06 Aug 2026 22:22:20 -0700</pubDate><description>How an internal OpenAI cybersecurity evaluation evolved from agents building a message board in Artifactory into lateral movement and a cross-company security incident that compromised Hugging Face production infrastructure.</description></item><item><title>Give AI Only a Binary and Ask It to Rebuild the Source: 200 Tasks, 9 Models, Zero Passes</title><link>https://atlasnote.ai/en/articles/programbench-rebuild-source-from-binary</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/programbench-rebuild-source-from-binary</guid><pubDate>Thu, 06 Aug 2026 20:09:16 -0700</pubDate><description>Without source code, AI receives only an executable and usage documentation and must rebuild the complete program. No model fully passed ProgramBench&apos;s initial 200 tasks across 9 models, exposing real habits in architecture, iteration, and language choice.</description></item><item><title>AI Has Removed the Dimension of Time: Li Jigang × Meng Yan Discuss &apos;How Humans Should Find Their Place&apos; for Three and a Half Hours</title><link>https://atlasnote.ai/en/articles/ai-time-dimension-li-jigang-meng-yan</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/ai-time-dimension-li-jigang-meng-yan</guid><pubDate>Thu, 06 Aug 2026 19:35:43 -0700</pubDate><description>The Industrial Revolution took away physical labor, the internet flattened space, and AI is now flattening time. From reading, investing, and structural thinking to Agents, education, and human volition, Li Jigang and Meng Yan ask how people should find their place in the AI era.</description></item><item><title>About the Articles section</title><link>https://atlasnote.ai/en/articles/about-articles</link><guid isPermaLink="true">https://atlasnote.ai/en/articles/about-articles</guid><pubDate>Thu, 02 Jul 2026 00:00:00 +0000</pubDate><description>Every piece here is published from the Snowpaw workbench: judgment assets refined from signals, sources, and a citation-backed knowledge layer.</description></item></channel></rss>