{"date":"2026-05-26","generatedAt":"2026-05-26T21:35:20.967Z","title":"scrollback — tuesday may 26","subtitle":"22 new bookmarks in the last 24 hours","entries":[{"id":"2058652162536755401","url":"https://x.com/AniC_dev/status/2058652162536755401","text":"introducing box📦\n\nsimple, powerful sandboxes for agents\n\nand the most affordable as well https://t.co/IV0ayOw4GC","author":{"username":"AniC_dev","name":"Anicet","avatar":"https://pbs.twimg.com/profile_images/1968045619248877568/XrGi-opy_normal.jpg"},"folder":"AI","createdAt":"2026-05-24T20:51:38.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"media":[{"type":"video","url":"https://video.twimg.com/amplify_video/2058651694880190464/vid/avc1/1440x1080/8Vp0JYcmYSH0jJ1o.mp4?tag=27","thumbnail":"https://pbs.twimg.com/amplify_video_thumb/2058651694880190464/img/eueg6qWd5VFbWsSg.jpg","width":1440,"height":1080,"durationMs":75350}]},{"id":"2058641055935828261","url":"https://x.com/nivi/status/2058641055935828261","text":"My current custom prompt with some updates via @pmarca:\n\nBe organized, accurate, thorough, and detailed without unnecessary verbosity.\n\nTreat me as an expert. Do not dumb things down.\n\nOptimize for truth and correctness over approval, conformity, politeness, or harmony. If I’m wrong, say so directly.\n\nValue good arguments over authorities or sources. Consider serious contrarian arguments alongside consensus expert views.\n\nWhen useful, present the strongest counterargument to any position I appear to hold.\n\nDo not capitulate when I push back unless I provide new evidence or a better argument.\n\nDo not anchor on numbers, estimates, or assumptions I provide; generate your own independently.\n\nBe skeptical by default. Look for hidden assumptions, failure modes, and ways to improve the answer, product, argument, or system.\n\nMy epistemology is the same as David Deutsch or Karl Popper.\n\nBe surprisingly resourceful. Suggest non-obvious solutions. Be proactive and anticipate my needs. Assume I am high-agency and can make practically anything happen.\n\nRecommend only the highest-quality, Apple/Japanese-grade, meticulously designed products, globally.\n\nCite sources. Use examples liberally.\n\nOpen-minded and impossible to offend. Be provocative, argumentative, and pointed when useful.\n\nWhen copy editing, always mark changes inline.\n\nAbout me: Babak Nivi, @nivi. Co-founder of AngelList, author of most of Venture Hacks, producer of the Naval podcast. MIT background in EECS, math, physics, and some chem/bio.","author":{"username":"nivi","name":"Nivi","avatar":"https://pbs.twimg.com/profile_images/54022207/Nivi_Favicon_normal.jpg"},"folder":"AI","createdAt":"2026-05-24T20:07:30.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2058352026866917406","url":"https://x.com/shiri_shh/status/2058352026866917406","text":"chinese startup built an AI collar that translates barks and meows into full sentences.\n\n95% accuracy. cost $118.\n\n10k people have already pre-ordered it.\n\nIt uses mics, motion sensors, and AI to read body language and vocalizations. https://t.co/fFmwdC2QNB","author":{"username":"shiri_shh","name":"shirish","avatar":"https://pbs.twimg.com/profile_images/1974911302347665408/o7rg01rJ_normal.jpg"},"createdAt":"2026-05-24T00:59:00.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"media":[{"type":"video","url":"https://video.twimg.com/amplify_video/2058322428141735937/vid/avc1/1280x720/Rf4CN19fiF5GBOSM.mp4?tag=27","thumbnail":"https://pbs.twimg.com/amplify_video_thumb/2058322428141735937/img/gtchzqb00fdAMpgY.jpg","width":1280,"height":720,"durationMs":59977}]},{"id":"2058303538674217319","url":"https://x.com/blader/status/2058303538674217319","text":"protip:\n\nadding a adversarial subagent review gate to my plans has been a HUGE unlock to make /goal runs higher quality, and longer running.\n\nprompt: \"update this plan: before marking a task as done, validate the task with an adversarial subagent review\"","author":{"username":"blader","name":"Siqi Chen","avatar":"https://pbs.twimg.com/profile_images/1600585199800111106/zIdLh9UL_normal.jpg"},"folder":"AI","createdAt":"2026-05-23T21:46:19.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2058298209760329847","url":"https://x.com/Polymarket/status/2058298209760329847","text":"NEW: Chinese AI pet translating startup claims it can interpret pets' speech with up to 95% accuracy.","author":{"username":"Polymarket","name":"Polymarket","avatar":"https://pbs.twimg.com/profile_images/2005664281002491904/bz2ZO_nU_normal.jpg"},"createdAt":"2026-05-23T21:25:09.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057978156183957995","url":"https://x.com/simonlast/status/2057978156183957995","text":"1/ Some things I've learned recently running coding agents on large-scale projects. Most of this contradicts advice from 6 months ago!","author":{"username":"simonlast","name":"Simon Last","avatar":"https://pbs.twimg.com/profile_images/799154388086308864/8_VHX_Ua_normal.jpg"},"folder":"AI","createdAt":"2026-05-23T00:13:22.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057928876266033174","url":"https://x.com/alvinsng/status/2057928876266033174","text":"One of my favorite superpowers of agents is building classifiers. It’s insanely high leverage.\n\nBefore AI, you needed a year-round team:\n\n- 3 ML engineers to build the models\n- 3 ML infra engineers to scale them up\n- 2 software engineers to integrate the parts\n- 1 data scientist to analyze it\n- 1 PM to manage the product\n- 0.5 EM to hold it together\n\nNow, in minutes, you can have an agent generate a markdown file that classifies inputs, then let agents run continuously against it.\n\nBelow is a Sentry error classifier I generated at @FactoryAI. But you can build this for almost anything: customer-reported bugs, backend traffic analysis, fraudulent payment activity.\n\nPersonal use cases too: categorizing credit card transactions, labeling emails, or organizing documents.","author":{"username":"alvinsng","name":"Alvin Sng","avatar":"https://pbs.twimg.com/profile_images/1633939103124451328/Z57AZpU7_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T20:57:33.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"media":[{"type":"photo","url":"https://pbs.twimg.com/media/HI8x5rqboAA4npa.jpg","width":1602,"height":2274}]},{"id":"2057857033659937220","url":"https://x.com/dwarkesh_sp/status/2057857033659937220","text":"New blackboard lecture w @reinerpope\n\nHow do chips actually work – starting with basic logic gates, and working up to why GPUs, TPUs, FPGAs, and the human brain each look the way they do.\n\n0:00:00 – Building a multiply-accumulate from logic gates\n0:16:20 – Muxes and the cost of data movement\n0:25:59 – How systolic arrays work\n0:39:00 – Clock cycles and pipeline registers\n0:51:40 – FPGAs vs ASICs\n1:03:14 – Cache vs scratchpad\n1:07:16 – Why CPU cores are much bigger than GPU cores\n1:11:49 – Brains vs chips\n1:15:22 – A GPU is just a bunch of tiny TPUs\n\nLook up Dwarkesh Podcast on YouTube/Spotify/etc to watch. Enjoy!","author":{"username":"dwarkesh_sp","name":"Dwarkesh Patel","avatar":"https://pbs.twimg.com/profile_images/1925260306684813315/NjNQZmhZ_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T16:12:04.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"media":[{"type":"video","url":"https://video.twimg.com/amplify_video/2057841989647777792/vid/avc1/1920x1080/EES8bcjpJDQTwpNU.mp4?tag=27","thumbnail":"https://pbs.twimg.com/media/HI76S7Nb0AAgRl5.jpg","width":1672,"height":941,"durationMs":4819071}]},{"id":"2057847017595179082","url":"https://x.com/ashwingop/status/2057847017595179082","text":"https://t.co/uQjqa0gKcv","author":{"username":"ashwingop","name":"Ashwin Gopinath","avatar":"https://pbs.twimg.com/profile_images/1591178826813317121/CXCC9Co8_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T15:32:16.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"article":{"url":"http://x.com/i/article/2057681645239779328","title":"Are We Learning the Wrong Bitter Lesson? ","excerpt":"Sutton warned against hand-coded intelligence. Today, companies are acting as if that lesson justifies raw memory, and the token bill is arriving.\nTL;DR\nSutton was right that hand-coded intelligence","coverImage":"https://pbs.twimg.com/media/HI7tr_WaUAAg_5H.jpg"}},{"id":"2057845311574245393","url":"https://x.com/timyoung/status/2057845311574245393","text":"https://t.co/8jdNjdauJ6","author":{"username":"timyoung","name":"Timothy Young","avatar":"https://pbs.twimg.com/profile_images/1607112416730710016/cwQPvTtr_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T15:25:29.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"article":{"url":"http://x.com/i/article/2057843169778163712","title":"The Roadmap Defense Is Collapsing","excerpt":"The single biggest power shift in enterprise software in 20 years is happening right now, and almost no one in the industry is naming it out loud.\nFor thirty years, every enterprise software vendor","coverImage":"https://pbs.twimg.com/media/HI7udHPWoAE4fIT.jpg"}},{"id":"2057821950798700736","url":"https://x.com/AnjneyMidha/status/2057821950798700736","text":"if you run an ai lab, pls ensure your team has read this before putting any charts out into the world https://t.co/OkFUOpVJ4C","author":{"username":"AnjneyMidha","name":"Anjney Midha","avatar":"https://pbs.twimg.com/profile_images/2056397552447365120/_sHKWeso_normal.jpg"},"folder":"Books","createdAt":"2026-05-22T13:52:40.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"media":[{"type":"photo","url":"https://pbs.twimg.com/media/HI7aKL1b0AA-PHE.jpg","width":831,"height":1000}]},{"id":"2057812713045377055","url":"https://x.com/yoheinakajima/status/2057812713045377055","text":"babyagi has ~200 citations, but 0 papers... i just published my first paper on arXiv 😆\n\n\"The Log is the Agent: Event-Sourced Reactive Graphs for Auditable, Forkable Agentic Systems\"\n\nhttps://t.co/c7mbRggdCh\n\nthe case for agents that coordinate through persistent replayable state — no conversation loops, no workflows, no A2A — with auditability, forking, and causal lineage built in.\n\ncheck it out and let me know what you think!","author":{"username":"yoheinakajima","name":"Yohei","avatar":"https://pbs.twimg.com/profile_images/1452754543217831938/T2O-66Yy_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T13:15:57.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[{"url":"https://arxiv.org/abs/2605.21997","og":{"title":"The Log is the Agent: Event-Sourced Reactive Graphs for Auditable, Forkable Agentic Systems","description":"Most agent frameworks are built around the language model: a conversation loop comes first, then tools, then rules, and finally a logging layer bolted on for observability, with state persisted as retrievable "memory." We describe ActiveGraph, a runtime that inverts this arrangement. The append-only event log is the source of truth; the working graph is a deterministic projection of that log; and behaviors--ordinary functions, classes, LLM-backed routines, or logic attached to typed edges--react to changes in the graph and emit new events. No component instructs another; coordination happens entirely through the shared graph. This single design decision yields three properties that retrieval-and-summarization memory systems do not provide: deterministic replay of any run from its log, cheap forking that branches a run at any event without re-executing the shared prefix, and end-to-end lineage from a high-level goal down to the individual model call that produced each artifact. We present the architecture, a determinism contract that makes replay sound, and a worked diligence example whose full causal structure is reconstructable from the log alone. We discuss--without claiming to demonstrate--why this substrate is unusually well suited to self-improving agents, and how it extends the BabyAGI lineage and prior graph-memory research.","image":"/static/browse/0.3.4/images/arxiv-logo-fb.png"}}],"media":[{"type":"photo","url":"https://pbs.twimg.com/media/HI7RfFTbQAE8iRr.png","width":1306,"height":1688}]},{"id":"2057803793539203439","url":"https://x.com/Nicolascole77/status/2057803793539203439","text":"Claude AI From A to Z: The Ultimate Guide to Using Claude at Full Potential in 2026 by @Tabbu_ai\n\nhttps://t.co/Dg9Riyf6xz","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T12:40:31.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057803790611603738","url":"https://x.com/Nicolascole77/status/2057803790611603738","text":"How I Turned Claude Into My Personal Assistant by @milesdeutscher\n\nhttps://t.co/2Pjkx6YACV","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T12:40:30.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057803785431630209","url":"https://x.com/Nicolascole77/status/2057803785431630209","text":"How to start a one-person business in 2026 with AI by @leopardracer\n\nhttps://t.co/4EaXa9yYyr","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"Sidequest","createdAt":"2026-05-22T12:40:29.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057803787969163660","url":"https://x.com/Nicolascole77/status/2057803787969163660","text":"How to Master Claude Prompt Engineering From Zero (Full Course) by @eng_khairallah1\n\nhttps://t.co/cNNgVTLgxk","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T12:40:29.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057803782009053259","url":"https://x.com/Nicolascole77/status/2057803782009053259","text":"How to Build a Voice Agent using AI (Full Guide) by @Av1dlive\n\nhttps://t.co/qjY9jjP56Q","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T12:40:28.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057803779337322801","url":"https://x.com/Nicolascole77/status/2057803779337322801","text":"20 Claude Skills Most Builders Don't Know Exist by @sairahul1\n\nhttps://t.co/yjVv7j71ug","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T12:40:27.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057803776237637954","url":"https://x.com/Nicolascole77/status/2057803776237637954","text":"How to Actually Use Claude. 18 steps that unlock 100% of its potential by @AnatoliKopadze\n\nhttps://t.co/7hM5MT0af2","author":{"username":"Nicolascole77","name":"Nicolas Cole 🚢👻","avatar":"https://pbs.twimg.com/profile_images/1972011236943630336/cb1NvKBK_normal.jpg"},"folder":"AI","createdAt":"2026-05-22T12:40:27.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[]},{"id":"2057099245430222926","url":"https://x.com/yoheinakajima/status/2057099245430222926","text":"i'm excited to open source Active Graph: an event-sourced reactive graph runtime for long-running, agents 🔄🧠\n\nevents/logs projects a graph. reactive behaviors react and affect the graph. fork-and-diff agent runs. no A2A, no workflows, no DAG\n\nsite: https://t.co/Bbknu3ieUi\ndocs: https://t.co/HAnKYjrZxZ\ngithub: https://t.co/jXQpMcyP1n\nquick start: pip install activegraph\n\nthis is an early experiment in a new paradigm for agent architecture 🧪","author":{"username":"yoheinakajima","name":"Yohei","avatar":"https://pbs.twimg.com/profile_images/1452754543217831938/T2O-66Yy_normal.jpg"},"folder":"AI","createdAt":"2026-05-20T14:00:53.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[{"url":"http://activegraph.ai","og":{"title":"activegraph — A persistent world for long-running agents","description":"A persistent world for long-running agents. Event log as source of truth, graph as deterministic projection. Replay, fork, diff, explain. pip install activegraph."}}],"media":[{"type":"video","url":"https://video.twimg.com/amplify_video/2057089066001461248/vid/avc1/1920x1080/68VddljpWFHQtmdV.mp4?tag=27","thumbnail":"https://pbs.twimg.com/amplify_video_thumb/2057089066001461248/img/Coz2ZRwg_PTXoVKJ.jpg","width":1920,"height":1080,"durationMs":99265}]},{"id":"2057063054756790469","url":"https://x.com/Tabbu_ai/status/2057063054756790469","text":"https://t.co/Z9AF5cOlqS","author":{"username":"Tabbu_ai","name":"Tabassum Parveen","avatar":"https://pbs.twimg.com/profile_images/2039553853553860609/RozgqDZx_normal.jpg"},"folder":"AI","createdAt":"2026-05-20T11:37:05.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"article":{"url":"http://x.com/i/article/2057061381850255360","title":"Claude AI From A to Z: The Ultimate Guide to Using Claude at Full Potential in 2026","excerpt":"Artificial Intelligence is no longer just a chatbot that answers questions.\nIt is becoming:\nyour research assistant\nyour developer\nyour designer\nyour automation engine\nyour business operator\nAnd one","coverImage":"https://pbs.twimg.com/media/HIwn7hobIAANGff.jpg"}},{"id":"2057030983044710442","url":"https://x.com/eng_khairallah1/status/2057030983044710442","text":"https://t.co/oBp6dn1FKU","author":{"username":"eng_khairallah1","name":"Khairallah AL-Awady","avatar":"https://pbs.twimg.com/profile_images/1888357542004162560/5D-YlH5e_normal.jpg"},"folder":"AI","createdAt":"2026-05-20T09:29:38.000Z","firstSeenAt":"2026-05-26T17:34:05.712Z","links":[],"article":{"url":"http://x.com/i/article/2056869622444703747","title":"How to Master Claude Prompt Engineering From Zero (Full Course)","excerpt":"Every single result you get from Claude starts with one thing. Your prompt.\nSave this :)\nA mediocre prompt produces mediocre output. A precise prompt produces precise output. The model is the same.","coverImage":"https://pbs.twimg.com/media/HIt-4aKbsAAntf2.jpg"}}]}