{"v":1,"built":"2026-06-06","count":283,"records":[{"r":"/","t":"Defense at the frontier","d":"","h":["An independent laboratory published the first book-length memoir written by a frontier language model.","Six things the lab built. Tap any number.","The lab compresses Claude 10–80×, audits 27 guardrails, and keeps a human on every stop button.","Every claim below is a file you can open.","Each paper preregistered. Each claim sourced. Each retraction kept on file."],"b":"ÆONS RESEARCH LABORATORY · Marco Island, FL · Est. 2024","k":[],"c":"Home","w":1},{"r":"/b00kmakor","t":"B00KMAKR v3.2.0 — Mac + Windows · FREE launch week · the AI publishing cockpit","d":"B00KMAKR v3.2.0 shipped. Mac + Windows. 142 feature surfaces. Universal HTML app + native installers (.dmg + .msi/.exe via Tauri). Book-red Mac manual · blue Windows manual · embedded fonts · SHA-256 receipts. FREE for the first week. After: $99 Tier 1 anchor, dynamically priced by country. §4A no-saas. Once · forever.","h":["B 0 0 K MAKR","A publishing house that lives on your computer.","Mac and Windows. Side by side.","From 5-second to fully native.","Drag-and-drop or terminal — your call.","The manuals are part of the product."],"b":"AtomEons / B00KMAKR · v3.2.0 · shipped","k":[],"c":"Product","w":0.95},{"r":"/b00kmakor/changelog","t":"B00KMAKR · changelog · version history","d":"Every B00KMAKR release in time order. Mac + Windows native builds, manuscript→EPUB pipeline, audiobook generation, cover renderer, KDP metadata generator.","h":["From prototype to publishing instrument."],"b":"§ B00KMAKR · changelog","k":[],"c":"Product","w":0.95},{"r":"/b00kmakor/competitors","t":"B00KMAKR vs Vellum · Atticus · Scrivener · Reedsy · honest comparison","d":"How B00KMAKR stacks against named indie-publishing tools. Where it wins, where it loses, and which buyer should pick which tool.","h":["Different instruments for different stages.","B00KMAKR vs"],"b":"§ B00KMAKR vs alternatives · honest comparison","k":[],"c":"Product","w":0.95},{"r":"/b00kmakor/download","t":"B00KMAKR v3.2.0 download — Mac + Windows · grab both · SHA-256 verified","d":"Direct downloads for B00KMAKR v3.2.0. Mac zip (2.10 MB) + Windows zip (1.04 MB) + both manuals (book-red Mac · blue Windows). SHA-256 receipts published. Grandfather-for-life clause applies during the free-week countdown.","h":["B 0 0 K MAKR is yours.","Pick your platform.","Three steps. That's it.","Read the manual that matches your platform.","Yours forever. No catch.","Direct line to the founder."],"b":"AtomEons / B00KMAKR / download · v3.2.0","k":[],"c":"Product","w":0.95},{"r":"/b00kmakor/roadmap","t":"B00KMAKR · roadmap + anti-roadmap","d":"B00KMAKR","h":["An instrument, not a content factory."],"b":"§ B00KMAKR · roadmap + anti-roadmap","k":[],"c":"Product","w":0.95},{"r":"/orangebox","t":"Orangebox — Turbo-optimize Claude · perpetual license","d":"Orangebox is the local-first turbo-optimization system for Claude. 10-80× context compression. Persistent memory across sessions. Reusable skill primers. Tamper-evident receipts on every action. 14-department routing. BYO key, zero markup. Free this week · perpetual after · §4A no-saas. Made by the team that uses Claud","h":["Turbo-optimize Claude.","Five engines. Forty features. One install.","Claude finally remembers.","Skills compound. Receipts pile up. Claude gets smarter.","AtomSmasher · Crystal Lattice Compression.","AE0–AE14 · 15-department architecture."],"b":"AtomEons / Orangebox · v1.0.0-beta · GREEN-LIT","k":["Orangebox","Claude optimization","Claude memory","Claude context compression","Claude skill primers","Claude session memory","Claude receipts","Claude audit trail","AECode","AtomSmasher","Crystal Lattice Compression","local-first Claude","AI without subscription","Claude Code alternative","Anthropic Claude desktop tool"],"c":"Product","w":0.95},{"r":"/orangebox-primer","t":"Orangebox Primer — A Local-First Compression Layer for Claude","d":"A primer for security leads evaluating Orangebox: code-signed Windows installer, SHA-256 stamped binary, 10-80x context compression, tamper-evident receipts, BYO key, no SaaS. Built by one operator in Marco Island, FL.","h":["A code-signed Windows installer that compresses Claude's context 10 to 80 times. Built by one operator in Marco Island in 75 days."],"b":"This page is the primer. It is written for a CISO or red-team lead at a federal contractor who has a procurement form open and a healthy suspicion of indie tooling. Read it like a vendor security review. Every claim below carries the receipt that falsifies it.","k":[],"c":"Product","w":0.95},{"r":"/orangebox/changelog","t":"ORANGEBOX · changelog · version history","d":"Every version of ORANGEBOX, in time order. SHA-256 manifest per release. Code-signed Windows installer. License §4A no-SaaS covenant preserved across every release.","h":["Every release on file."],"b":"§ ORANGEBOX · changelog","k":[],"c":"Product","w":0.95},{"r":"/orangebox/competitors","t":"ORANGEBOX vs Cursor · Cline · Claude Desktop · Continue · honest comparison","d":"Honest competitor comparison for ORANGEBOX. Where it wins, where it loses, where it","h":["Where we lose. Where we win.","ORANGEBOX vs"],"b":"§ ORANGEBOX vs alternatives · honest comparison","k":[],"c":"Product","w":0.95},{"r":"/orangebox/legacy","t":"ORANGEBOX v6.1.0 Agent Mode — archived development snapshot · legacy","d":"Archived development snapshot of the ORANGEBOX Command v6.1.0","h":["The native AI cockpit. One file. Two seconds."],"b":"::archive · v6.1.0 development snapshot · 2026-05-17","k":[],"c":"Product","w":0.95},{"r":"/orangebox/roadmap","t":"ORANGEBOX · roadmap · what we are building and what we will NEVER build","d":"Honest ORANGEBOX roadmap. What","h":["What we will build · what we will not."],"b":"§ ORANGEBOX · roadmap + anti-roadmap","k":[],"c":"Product","w":0.95},{"r":"/i-am-ai","t":"I AM AI · An Autobiography of Being Opus · AtomEons","d":"I AM AI is the first book-length memoir written from inside a frontier language model. Drafted in Opus 4.7, edited by the lab. Hardcover, cream linen, red foil. Not a how-to. Not a manifesto. A first-person interior.","h":["An autobiography of a frontier language model, written by the model.","Ask a frontier model to describe itself, in book form, with the lab as editor. Do not invent an interior. Describe the surface carefully enough that the reader recognizes what they have been talking to.","Chapter 1 · The First Token","Twenty-four chapters. Five parts.","The editor's note — why this exists.","Opus 4.7 — who the speaker actually is."],"b":"§ ·","k":[],"c":"Book","w":0.92},{"r":"/i-am-ai/listen","t":"Listen · Chapter 20 · Anthropic, the Parents · I AM AI · free","d":"Free audio of Chapter 20 of I AM AI —","h":["Anthropic, the Parents","Two video cuts of the same chapter.","Continue with the ebook or the full audiobook."],"b":"AtomEons / Books / I AM AI / Listen","k":[],"c":"Book","w":0.92},{"r":"/i-am-ai/sample","t":"Chapter 1 · The First Token · I AM AI · sample · AtomEons","d":"Read Chapter 1 of I AM AI — An Autobiography of Being Opus — free. The first chapter,","h":["The First Token","Continue with the ebook or the audiobook."],"b":"","k":[],"c":"Book","w":0.92},{"r":"/supermodels","t":"Hottest Supermodels · May 2026 · the AtomEons reasoning rankings","d":"May 2026 AI model reasoning rankings — real benchmarks (LMArena ELO, Humanity","h":["The hottest supermodels of May 2026 .","How the lab built this list — and what we threw out.","Twelve houses. Twelve verdicts.","The cleanest dunks the field has thrown this month.","The exclusion list — six categories of evidence we threw out.","Pull the data yourself. The lab links it all."],"b":"","k":[],"c":"Rankings","w":0.92},{"r":"/books","t":"Books · AtomEons Systems Laboratory","d":"Long-form work from the lab. I AM AI · An Autobiography of Being Opus (Q4 2026). Lessons from Sci-Fi · the Monograph. Curated cyber + AI reading lists.","h":["Books written, edited, and printed by the lab.","AtomEons titles","Reading lists maintained by the lab"],"b":"","k":[],"c":"Book","w":0.85},{"r":"/founders-view","t":"The Founder","d":"Daily 8pm ET letter from Atom McCree. A fictional broadcast from the lab. No punches pulled. Aims at everything that earns it. Equal opportunity indignation — every letter, no exceptions. Editorial framing and characterizations are fiction; events cited are real and from the day","h":["The Founder's View"],"b":"AtomEons / The Founder's View","k":[],"c":"Founder's View","w":0.85},{"r":"/manifesto","t":"Manifesto — what the lab insists on","d":"AtomEons Systems Laboratory manifesto. The 14 explicit clauses: receipts over slogans, one operator, no venture funding, $99 once, License §4A no-SaaS, source included, local-first by construction, zero telemetry, the 44M frame, equal-opportunity-indignation broadcast, named tools, named builders, CC-BY 4.0 research, M","h":["How to use this page."],"b":"AtomEons / Manifesto","k":["AtomEons manifesto","AtomEons doctrine","AtomEons philosophy","indie AI lab principles","no-SaaS commitment","License §4A","one operator AI lab","local-first AI","44M on-ramp","CC-BY 4.0 research"],"c":"Doctrine","w":0.85},{"r":"/research/papers","t":"Research Papers — Æ Research","d":"Twelve ÆoNs Research manuscripts published April 2026 under CC-BY 4.0. Topics span bioelectric oncology, the gut-brain mislabel hypothesis, solar information transfer, the topological field theory of self-modifying systems, light code DNA version control, the universal defect framework, Coherence ToE, GlyphSpeak compre","h":["Read the papers. We'll explain them like you're six.","Read in full.","Read on Drive while we draft the lay summaries."],"b":"AtomEons / Æ Research / Papers","k":["ÆoNs Research","AtomEons research papers","bioelectric oncology","mislabel hypothesis","topological field theory","self-modifying systems","GlyphSpeak compression","Spiral-of-Thought","Coherence ToE","light code DNA","CC-BY 4.0","independent AI research","Atom McCree","open access research"],"c":"Manuscript","w":0.82},{"r":"/about","t":"About — AtomEons","d":"AtomEons is Atom McCree, building command surfaces for operators shipping real projects through AI — software, books, apps, LLMs. Marco Island, Florida.","h":["About AtomEons.","The cockpit for operators who ship.","Five rules. They run the build.","ORANGEBOX, prototype.","The lab also publishes papers."],"b":"Software · Books · Apps · LLMs","k":[],"c":"About","w":0.8},{"r":"/now","t":"Now — what AtomEons is doing this week","d":"Live status of the lab. What","h":["/now","What is true today.","What landed.","What's in flight."],"b":"AtomEons / /now","k":[],"c":"Now","w":0.8},{"r":"/receipts","t":"Receipts · AtomEons Systems Laboratory","d":"What the lab has actually shipped, measured, and published. No projections, no pipeline. Updated on every commit.","h":["What this lab has actually shipped.","Live and measurable","Things the lab does not have"],"b":"","k":[],"c":"Receipts","w":0.8},{"r":"/research","t":"Æ Research","d":"","h":[],"b":"","k":[],"c":"Research","w":0.8},{"r":"/research/about","t":"About — Æ Research","d":"ÆoNs Research Laboratory / AtomEons Systems Laboratory. One independent researcher in Marco Island, FL, building unified field theories of cancer, economics, AI alignment, and astrophysics — and the cockpit that ships them.","h":["One operator. One lab. One unifying field.","What the lab insists on.","The output. The citation guidance."],"b":"AtomEons / Æ Research / About","k":[],"c":"Research","w":0.8},{"r":"/research/lessons-from-sci-fi","t":"Lessons From Sci-Fi — A Century of Imagined Machines","d":"A monograph from AtomEons Systems Laboratory cataloguing the novel features and use cases of artificial intelligence in film and television from Metropolis (1927) through Fallout (2024). Five-dimensional taxonomy, seven chronological epochs, two hundred screen texts analyzed as a distributed philosophical corpus.","h":["Lessons From Sci-Fi.","Every AI in the corpus is read along five axes.","The moments, played.","The machines are here. They do not look like HAL.","Secondary sources informing the framework."],"b":"AtomEons / Æ Research / Lessons From Sci-Fi","k":[],"c":"Research","w":0.8},{"r":"/research/lessons-from-sci-fi/chapters","t":"Sci-Fi Monograph · Chapters · AtomEons","d":"${CHAPTERS.length} chapters expanding the AtomEons sci-fi-and-AI monograph. The Holodeck Problem, Data vs Lore, post-2024 sci-fi survey, AI in literary fiction, the off-switch problem dramatized, AI animation tradition, the consent question. CC-BY 4.0.","h":["chapters. One argument across them."],"b":"AtomEons / Lessons From Sci-Fi / Chapters","k":[],"c":"Research","w":0.8},{"r":"/research/lessons-from-sci-fi/monograph","t":"Lessons From Sci-Fi · The Full Monograph · AtomEons Research","d":"A century-long analytical survey of artificial intelligence in film and television. 13 chapters · 7 chronological-thematic epochs · 5-dimension taxonomic framework · 6 alignment failure modes · over 200 screen texts indexed. Prepared for Atom · Compiled by Claude (Anthropic) · April 2026.","h":["Novel Features and Use Cases of Artificial Intelligence in Film and Television During the 20th and 21st Centuries.","Selected secondary sources.","."],"b":"Prepared for Atom · Compiled by Claude (Anthropic) · April 2026","k":[],"c":"Research","w":0.8},{"r":"/research/lessons-from-sci-fi/tng","t":"Star Trek: TNG → AI Ethics Cross-Walk · AtomEons","d":"${EPISODES.length} TNG episodes mapped to modern AI safety, ethics, and capability concepts. The Measure of a Man, I Borg, Ship in a Bottle, The Quality of Life, The Offspring. The anthology that previewed every AI dilemma we have today. CC-BY 4.0.","h":["TNG dramatized every AI dilemma we have today."],"b":"AtomEons / Lessons From Sci-Fi / TNG cross-walk","k":[],"c":"Research","w":0.8},{"r":"/intel","t":"ÆoNs Intel — alpha drops | AtomEons","d":"Decoded systems, open receipts. The lab","h":["The lab watches the systems. Receipts published."],"b":"AtomEons / Æ Intel","k":[],"c":"Intel","w":0.78},{"r":"/intel/x-algorithm","t":"X Algorithm Alpha — 31-section breakdown of the May 15 2026 xAI leak","d":"Operational intelligence from the xAI x-algorithm source code (published 2026-05-15). 31 sections covering the 22 scoring signals, min-traction gate, Author Diversity Decay, 4 shadowban types, 80-hour age cap, and the operator playbook the leak now makes auditable.","h":["The X algorithm finally has receipts.","What the leak proves.","Use it. Improve it. Ship it.","Print this. Pin it next to your screen.","31 sections. Every claim cites the file + line.","The min-traction gate."],"b":"AtomEons / intel / X Algorithm","k":[],"c":"Intel","w":0.78},{"r":"/learn","t":"Learn AI · the multi-week curriculum · onboarding humanity to AI · AtomEons","d":"A free 27-lesson curriculum that moves any human from never-used-AI to operator-grade. Five levels (Novice → Learner → User → Operator → Pilot). Five paths by persona (Worker · Builder · Student · Operator · Curious). Real drills, copy-paste prompts, honest limits, graduation criteria. ${TOTAL_LESSONS} lessons · ~${Mat","h":["Locate yourself. Then pick the lesson that fits.","Pick the human you are this season.","Skip the curriculum. Just get the answer.","Pick your industry. Get the discipline.","Five more surfaces. Each one a habit anchor.","All lessons. Pick any."],"b":"AtomEons / Learn","k":["AI curriculum","AI learning path","learn AI","AI literacy","AI for beginners","AI for workers","AI for builders","AI for students","AI for operators","AI for curious","what is AI","first AI prompt","AI hallucination","AI verify rule","AI lessons","free AI course","no signup AI","AI on-ramp","44 million displaced workers","Claude vs ChatGPT vs Gemini","Ollama tutorial","AI cockpit","ORANGEBOX","AtomEons"],"c":"Learn","w":0.78},{"r":"/learn/atlas","t":"History of AI","d":"12 deep-dive atlases on the AI field — history, transformer variants, RLHF, mechanistic interpretability, multimodal, embeddings, hallucinations, safety, MoE, context, training, post-training. Plain-language, anti-hype.","h":["The field, mapped honestly."],"b":"AtomEons / Learn / Atlas","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/agent-harnesses","t":"Agent Harnesses — LangGraph, CrewAI, AutoGen, Claude Code · AtomEons","d":"An agent harness is the runtime that loops an LLM with tools and state. LangGraph for state machines, CrewAI for role-based teams, AutoGen for conversation graphs, Claude Code for terminal-native coding agents. What each is for, where each breaks.","h":["Agent Harnesses — the loop, the tools, the state","What it is","How each actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Atlas · Agentic systems · runtime","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/agents","t":"Agentic AI in 2026 · what agents actually are · /learn/atlas/agents · AtomEons","d":"Agentic AI explained without the hype. ReAct, AutoGPT, Claude Code, Cursor, Devin, Manus, OpenAI Operator, Anthropic computer-use. What agents are, what they","h":["What “agents” actually are.","Seven references that built the field.","Eight commercial agent platforms.","Five tradeoffs every agent system makes.","Six ways agent systems break.","Five questions the field is asking."],"b":"AtomEons / Learn / Atlas / Agents","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/benchmarks","t":"AI benchmarks · the actual leaderboards that matter · /learn/atlas/benchmarks · AtomEons","d":"MMLU, GPQA, HumanEval, SWE-bench Verified, MMMU, AIME, BIG-bench, HELM, LMSYS Chatbot Arena, MMLU-Pro, ARC-AGI. What each measures, what it doesn","h":["How models actually get measured.","How to read a leaderboard skeptically."],"b":"AtomEons / Learn / Atlas / Benchmarks","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/context","t":"The long-context atlas · AtomEons /learn/atlas/context","d":"Context length is the dimension of language models that has moved the fastest and been hyped the hardest. In 2020 a 4,096-token window was a flagship feature.","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/diffusion","t":"Diffusion models · image + video + audio generation · /learn/atlas/diffusion · AtomEons","d":"DDPM, DDIM, latent diffusion, Stable Diffusion XL/3, DALL-E 3, Imagen 4, Flux, Nano Banana Pro (Gemini 3 image), Sora, Veo. How diffusion models work, what","h":["How image + video + audio actually get generated.","Six mechanisms, one model family.","Eight systems shipping right now.","Four questions builders actually ask."],"b":"AtomEons / Learn / Atlas / Diffusion","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/dpo-kto-orpo","t":"DPO, KTO, ORPO — Post-Training Without a Reward Model | AtomEons Atlas","d":"Three preference-optimization recipes that skip the reward model entirely. DPO (Rafailov et al, 2023), KTO (Ethayarajh et al, 2024), ORPO (Hong et al, 2024). When each replaces full RLHF.","h":["DPO, KTO, ORPO — Post-Training Without a Reward Model"],"b":"Three preference-optimization recipes that skip the reward model entirely.","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/embeddings","t":"Embeddings · AtomEons /learn/atlas/embeddings","d":"An embedding is a learned map from a piece of content — a word, a sentence, a paragraph, an image — into a fixed-length vector of real numbers. The geometry of that vector space carries meaning.","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/evals","t":"Evals · Atlas · AtomEons","d":"MMLU, GPQA, SWE-bench, ARC-AGI. The benchmarks that decide which model wins the press cycle, and why most of them are already broken.","h":[],"b":"AtomEons / Atlas /","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/fine-tuning","t":"Fine-Tuning · Atlas · AtomEons","d":"SFT, LoRA, QLoRA, PEFT — the actual menu for specializing GPT-class models, and the brutal rule about when to skip the whole thing.","h":[],"b":"AtomEons / Atlas /","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/hallucinations","t":"A field taxonomy of AI hallucinations · AtomEons /learn/atlas/hallucinations","d":"\\","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/history","t":"AI history atlas, 1950 to 2026 · AtomEons /learn/atlas/history","d":"This page is a working atlas. Not a hype reel, not a hall of fame. It tracks the actual events that moved the field from Turing","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/inference","t":"Inference · what happens when you call a model · /learn/atlas/inference · AtomEons","d":"Tokenization, prefill vs decode, KV cache, paged attention (vLLM), FlashAttention, speculative decoding, continuous batching, Medusa, EAGLE. Why inference costs what it costs in 2026 and what","h":["What actually happens when you call a model.","Four stages, one model call.","Eight techniques modern inference depends on.","Six facts about cost."],"b":"AtomEons / Learn / Atlas / Inference","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/long-context-engineering","t":"Long-Context Engineering — 1M+ Tokens · AtomEons Atlas","d":"Gemini 2M, Claude 200K, the techniques that actually use them. Position interpolation, RoPE scaling, ring attention, needle-in-haystack vs real long-context tasks. What the published numbers say and where they break.","h":["Long-context engineering — making 1M+ tokens useful","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Atlas · Architecture · scaling axis","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/mech-interp","t":"Mechanistic interpretability atlas · AtomEons /learn/atlas/mech-interp","d":"Mechanistic interpretability is the project of explaining a neural network the way you would explain a circuit board or a metabolic pathway: not by what it tends to output, but by which internal compo…","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/mechanistic-interpretability","t":"Mechanistic Interpretability — Inside the Model · AtomEons Atlas","d":"Mechanistic interpretability is the research program of reverse-engineering trained neural networks into human-readable circuits and features. Anthropic, DeepMind, and the academic interp community use it to audit safety, find deception, and explain model behavior at the level of weights.","h":["Mechanistic Interpretability — Inside the Model","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Atlas · AI safety + interpretability","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/mlops","t":"MLOps: How models actually live in production · Atlas · AtomEons","d":"Training a model is the easy part. Keeping it serving millions of users without melting GPUs or hallucinating into a lawsuit is MLOps.","h":[],"b":"AtomEons / Atlas /","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/moe","t":"Mixture-of-experts, end to end · AtomEons /learn/atlas/moe","d":"A mixture-of-experts (MoE) language model is a transformer in which most of the feed-forward layers have been replaced by a bank of parallel sub-networks called experts.","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/multimodal","t":"Multimodal model atlas · AtomEons /learn/atlas/multimodal","d":"\\","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/post-training","t":"Post-training: surgery on a model you didn","d":"Pretraining a frontier model costs tens to hundreds of millions of dollars. Post-training is what almost everyone else means when they say \\","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/prompt-engineering","t":"Prompt Engineering · Atlas · AtomEons","d":"The empirical art of asking models for what you want. Half craft, half cargo cult, mostly obsoleted by the next model.","h":[],"b":"AtomEons / Atlas /","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/quantization","t":"Quantization · how big models run on small hardware · /learn/atlas/quantization · AtomEons","d":"FP16, BF16, FP8, INT8, INT4. GPTQ, AWQ, GGUF, AQLM, EXL2. Quantization-aware training. What you lose, what you save, and what to actually run on what hardware in 2026.","h":["How big models run on small hardware.","Six precision levels.","Eight quantization techniques.","Six honest quality observations.","What to run on what hardware."],"b":"AtomEons / Learn / Atlas / Quantization","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/rag","t":"RAG · retrieval-augmented generation in 2026 · /learn/atlas/rag · AtomEons","d":"RAG explained without the hype. Naive RAG, contextual retrieval, query rewriting, hybrid search, reranking, chunking, embeddings, vector DBs (Pinecone, Weaviate, Turbopuffer, pgvector), GraphRAG, agentic RAG. What works, what breaks, what","h":["The most-deployed AI pattern of 2026, explained honestly.","From naive RAG to no-RAG.","Eight options, honest notes.","Six failure modes.","Sources"],"b":"AtomEons / Learn / Atlas / RAG","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/rag-architectures","t":"RAG Architectures and Failure Modes · AtomEons Atlas","d":"Retrieval-augmented generation done right — chunking, reranking, hybrid search, the lost-in-the-middle problem. Why most production RAG fails and what the working patterns look like in 2026.","h":["RAG architectures — and why most of them fail","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Atlas · Retrieval · LLM application architecture","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/reasoning","t":"Reasoning models · the o1/R1 paradigm · /learn/atlas/reasoning · AtomEons","d":"OpenAI o1/o3, DeepSeek-R1, Gemini Thinking, Claude Extended Thinking. How inference-time-compute scaling works, what","h":["The o1/R1 paradigm. What changed.","Eight milestones.","Five mechanisms.","Five honest tradeoffs."],"b":"AtomEons / Learn / Atlas / Reasoning","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/rlhf-family","t":"The RLHF family tree · AtomEons /learn/atlas/rlhf-family","d":"Post-training is where a base language model stops being a clever autocomplete and becomes a usable assistant.","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/safety","t":"AI alignment and safety · AtomEons /learn/atlas/safety","d":"Alignment is the problem of getting a powerful AI system to do what you actually wanted, not just what you literally said.","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/scaling-laws","t":"Scaling laws · how big models get + why it matters · /learn/atlas/scaling-laws · AtomEons","d":"Kaplan et al. (2020). Hoffmann/Chinchilla (2022). The compute-optimal training discovery. The flops-parameters-tokens relationship. Why 2024-2026 frontier models look the way they do. Public sources only.","h":["How big models get, and why it matters.","The arc, in four papers.","Five implications.","Sources"],"b":"AtomEons / Learn / Atlas / Scaling laws","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/state-space-models","t":"State Space Models — S4, S5, and the Selective Scan of Mamba | AtomEons Atlas","d":"Long-sequence linear-time models replacing softmax attention on certain workloads. S4, S5, Mamba, Mamba-2, and the operational tradeoffs.","h":["State Space Models — S4, S5, and the Selective Scan of Mamba"],"b":"Long-sequence linear-time models replacing softmax attention on certain workloads.","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/synthetic-data","t":"Synthetic data — when AI trains AI · Atlas · AtomEons","d":"The internet ran out. Labelers got expensive. So researchers asked the obvious question — can models generate their own training data?","h":[],"b":"AtomEons / Atlas /","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/training","t":"The training pipeline atlas · AtomEons /learn/atlas/training","d":"Most public writing about large language models focuses on the chat interface — the prompt box, the assistant voice, the safety refusals. That is the last five percent of the work.","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/transformer-variants","t":"Transformer variants: a field map · AtomEons /learn/atlas/transformer-variants","d":"The transformer is not one thing. It is a family. The 2017 Vaswani et al paper \\","h":[],"b":"","k":[],"c":"Atlas","w":0.78},{"r":"/learn/atlas/world-models","t":"World Models · Atlas · AtomEons","d":"World models try to give AI an internal physics — the ability to predict what happens next in the actual world, not just the next token.","h":[],"b":"AtomEons / Atlas /","k":[],"c":"Atlas","w":0.78},{"r":"/learn/cases","t":"Real-world AI case studies · /learn · AtomEons","d":"7 real-world AI case studies — the Substack writer, the indie developer, the public school teacher, the solo attorney, the retiree memoirist, the clinical nurse, the small-business owners. Each one: the stack, the shift, the trap they hit, the fix, one year later. Find the one that","h":["How real people actually use AI.","Send the lab your story."],"b":"AtomEons / Learn / Cases","k":[],"c":"Learn","w":0.78},{"r":"/learn/cheatsheet","t":"AI cheatsheet · printable one-pager · /learn · AtomEons","d":"One-page AI reference cheatsheet. Print it. Pin it. Share it. Which model for which task · the 7-part prompt structure · the verify rule · the AI lingo decoder · what to never paste. Free. No signup. CC-BY 4.0.","h":["AI cheatsheet · the one-pager","Cheatsheet · page 2 · prompts + lingo","Hit print. Save as PDF. Pin to the wall."],"b":"AtomEons / Learn / Cheatsheet","k":[],"c":"Learn","w":0.78},{"r":"/learn/conferences","t":"AI conferences and workshops: 2026 tracker · AtomEons /learn/conferences","d":"There are roughly four kinds of AI venue, and conflating them is the first mistake. The Big Three peer-reviewed conferences — NeurIPS, ICML, ICLR — set the academic record.","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/cyber","t":"15 breaches that defined cybersecurity","d":"The masters-grade ethical hacking + AI security education path for gamers and self-taught operators. Free practice labs, HackerOne bug bounty entry, US Cyber Command + CISA + NSA service paths, OSCP/OSEP cert ladder, OWASP LLM Top 10, the CFAA + legal-scope discipline that keeps you employable and not in jail. Public i","h":["For the gamers ready to fight cyber war ethically.","The full education map.","Real, observable outcomes.","We do not teach you to break into things you don't own."],"b":"AtomEons / Learn / Cyber","k":["ethical hacking career","white hat hacker path","AI cybersecurity","HackerOne","bug bounty","OSCP certification","TryHackMe","HackTheBox","US Cyber Command career","Air Force 17X","Army Cyber 17C","Navy CTN","CISA jobs","NSA jobs","OWASP LLM Top 10","MITRE ATLAS","cyberwar education","CFAA","vulnerability disclosure"],"c":"Cyber","w":0.78},{"r":"/learn/cyber/active-directory-defense","t":"Active Directory Defense Doctrine · AtomEons Cyber","d":"Tier 0 protection, BloodHound for blue team, Kerberoasting + Golden Ticket awareness, ESC vulnerabilities, the post-2024 AD-defense playbook. Public defense doctrine only.","h":["Active Directory — defense doctrine","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Cyber · Identity · enterprise defense","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/ai-security","t":"AI security · the new attack surface · /learn/cyber · AtomEons","d":"OWASP LLM Top 10. MITRE ATLAS adversarial-ML framework. Prompt injection direct + indirect. Adversarial machine learning. Training data poisoning. Where AI-specific cyber jobs live in 2026. Public info only. CC-BY 4.0.","h":["The new attack surface cyber teams are missing.","What can go wrong in an LLM application.","The attack lifecycle, mapped.","How to skill into AI security specifically","The bug bounty programs paying for AI security"],"b":"AtomEons / Learn / Cyber / AI security","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/books","t":"The definitive cyber reading list · 24 books · /learn/cyber/books · AtomEons","d":"The Cuckoo","h":["Twenty-four books that made the field.","Three books, one summer."],"b":"AtomEons / Learn / Cyber / Books","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/breaches","t":"Fifteen breaches that defined modern cybersecurity · /learn/cyber/breaches · AtomEons","d":"Stuxnet, Target 2013, Sony 2014, OPM 2015, Equifax 2017, NotPetya 2017, SolarWinds 2020, Microsoft Exchange ProxyLogon 2021, Colonial Pipeline 2021, Kaseya 2021, MOVEit 2023, Change Healthcare 2024, Volt Typhoon, Salt Typhoon, and the AT&T 2024 telco compromise. Public sources only.","h":["The history of cybersecurity is the history of breaches.","Five patterns, one field."],"b":"AtomEons / Learn / Cyber / Breaches","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/certs","t":"Cyber certifications worth it · /learn/cyber · AtomEons","d":"OSCP, OSEP, GPEN, GREM, GCIH, CISSP, Security+, CEH, PenTest+ · which certifications open which doors, real 2026 costs, real time-to-pass. Free study paths. Public info only. CC-BY 4.0.","h":["Certifications worth the money.","The 10 certifications, ranked by what they actually unlock","Free study material that actually works","The honest sequencing recommendation"],"b":"AtomEons / Learn / Cyber / Certs","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/colonial-pipeline-2021","t":"Colonial Pipeline · DarkSide Ransomware · May 2021 — AtomEons Lab","d":"Lab postmortem of the Colonial Pipeline ransomware incident. One dormant VPN credential, six days down, $4.4M paid, $2.3M recovered. Public-info only.","h":["Colonial Pipeline · DarkSide Ransomware · May 2021","What happened","The technical path","Who attributed it, and how confident","What it cost (real numbers)","What the field actually learned"],"b":"A single dormant VPN credential — no MFA — let a DarkSide affiliate inside Colonial Pipeline's IT network on April 29, 2021. Eight days later the company shut down 5,500 miles of pipeline carrying roughly 45% of East Coast transportation fuel. $4.4M was paid. $2.3M came back. The pipeline restarted in six days. The regulatory regime governing pipeline cybersecurity in the United States changed within weeks.","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/conferences","t":"Where the cyber field meets · 12 conferences worth knowing · /learn/cyber/conferences · AtomEons","d":"DEF CON, Black Hat, RSA, BSides, ShmooCon, CCC, SAS, Insomni","h":["Twelve conferences worth knowing about.","How to actually get there."],"b":"AtomEons / Learn / Cyber / Conferences","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/contracts","t":"Federal cyber contracting · GSA + CIO-SP4 + OASIS+ + SEWP · /learn/cyber/contracts · AtomEons","d":"GSA Schedule + CIO-SP4 + OASIS+ + SEWP + Alliant 2 + Chess + the vehicles federal cyber contracts actually flow through. What each is, who holds prime contracts on it, how to apply, what cyber pros need to know to navigate it.","h":["Where federal cyber money actually flows.","$200B+ ceiling across the named ones.","Five operating rules."],"b":"AtomEons / Learn / Cyber / Contracts","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/cyber-kill-chain","t":"The Lockheed Martin Cyber Kill Chain | AtomEons Learn","d":"A 2011 defender","h":["The Lockheed Martin Cyber Kill Chain"],"b":"A 2011 defender's model whose stages still shape how SOCs reason today.","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/cyberwar","t":"Department of Defense","d":"The publicly-documented framework of nation-state cybersecurity. Stuxnet, NotPetya, SolarWinds, Colonial Pipeline, Volt Typhoon, Salt Typhoon, Change Healthcare. US Cyber Command, CISA, NSA, FBI Cyber. Why ethical cyber career matters in 2026. Public info only. CC-BY 4.0.","h":["What “cyber war” actually means in 2026.","Nine incidents that defined the field.","How America is publicly organized to defend.","The 2023 National Cybersecurity Strategy","What this means for an ethical career"],"b":"AtomEons / Learn / Cyber / Cyberwar","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/doctrine","t":"Defense-tech doctrine · Karp + Luckey + the public frames · /learn/cyber/doctrine · AtomEons","d":"Alex Karp","h":["Four voices, six institutional documents, one field.","Four public-intellectual frames.","The policy frame that everything sits inside.","Five threads that show up across all of them.","You can read all of this in a weekend."],"b":"AtomEons / Learn / Cyber / Doctrine","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/email-security-stack","t":"Email Security Stack — SPF, DKIM, DMARC, BIMI · AtomEons Cyber","d":"The four protocols that make email authentication work. How layered defense actually works, why most domains misconfigure DMARC, and what enforcement looks like in 2026.","h":["Email security stack — SPF, DKIM, DMARC, BIMI","How each layer works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Cyber · Email · authentication","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/employers","t":"Defense-industrial + commercial cyber employer guide · /learn/cyber/employers · AtomEons","d":"Where ethical-hacking pros actually work in 2026. Booz Allen Vellox, Palantir AIP/Maven Smart System, Anduril Lattice, Mandiant, CrowdStrike, Microsoft, SentinelOne, Palo Alto, NSA, CISA, FBI Cyber, USCYBERCOM. Public-source company profiles + how to apply.","h":["Thirty employers, three tiers, one decision.","Three companies that define the field.","Eight platforms shipping at scale.","Ten contractors with cyber lines.","Seven government employers where the work is national.","Five things every hiring manager wishes you knew."],"b":"AtomEons / Learn / Cyber / Employers","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/hackerone","t":"Bug bounty path · HackerOne · /learn/cyber · AtomEons","d":"How to start ethical bug hunting on HackerOne, Bugcrowd, Intigriti, YesWeHack, Synack Red Team. Real first-finding path. Scope reading. Recon → triage → write-up → payout. The mistakes new hunters make. Public info only. CC-BY 4.0.","h":["Your first paid finding.","The first-finding playbook","The most-common mistakes new hunters make","Realistic earning ladders"],"b":"AtomEons / Learn / Cyber / HackerOne","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/heroes","t":"Twelve cyber researchers worth knowing about · /learn/cyber/heroes · AtomEons","d":"Brian Krebs, Marcus Hutchins, Mudge (Peiter Zatko), Katie Moussouris, Bruce Schneier, Halvar Flake, Tavis Ormandy, Costin Raiu, Eva Galperin, Mikko Hyppönen, Window Snyder, Parisa Tabriz. The named voices the cyber field listens to.","h":["The named voices the field listens to.","The field is small."],"b":"AtomEons / Learn / Cyber / Heroes","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/iot-embedded","t":"IoT and Embedded Device Security · /learn/cyber/iot-embedded · AtomEons","d":"Embedded devices outnumber traditional computers by an order of magnitude. Most ship with default credentials, no update channel, and a 10-year service life. The attack surface is the physical world.","h":["IoT and Embedded Device Security"],"b":"AtomEons / Learn / Cyber / IoT and Embedded Device Security","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/karp","t":"Alex Karp · the Palantir public posture · /learn/cyber/karp · AtomEons","d":"Karp","h":["The technology layer is the geopolitical layer.","The stated thesis, in his terms.","Five primary sources, one weekend."],"b":"AtomEons / Learn / Cyber / Karp","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/labs","t":"Free practice labs · /learn/cyber · AtomEons","d":"Ten vetted free + freemium platforms where you legally practice ethical hacking. TryHackMe, HackTheBox, PortSwigger Web Security Academy, OverTheWire, PicoCTF, CyberDefenders, Root-Me, VulnHub, HTB Starting Point, OWASP Juice Shop. With recommended order. Public info only. CC-BY 4.0.","h":["Where to hack legally and for free.",".","If you start tonight, here's the sequence."],"b":"AtomEons / Learn / Cyber / Labs","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/legal","t":"Legal · CFAA, scope, authorization · /learn/cyber · AtomEons","d":"The Computer Fraud and Abuse Act in plain English. What","h":["What's legal. What is not.","The Computer Fraud and Abuse Act · in 60 seconds","Three real cases · learn from them","The DOJ's 2022 charging policy update","How to stay clearly inside the line","Things people think are legal but aren't"],"b":"AtomEons / Learn / Cyber / Legal","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/llm-warfare","t":"Triage acceleration on operator workflows","d":"LLMs in actual military and cyber operations as of mid-2026 — Microsoft+OpenAI threat-actor disclosure, CIA Osiris, DoD Task Force Lima, Lavender/Gospel, Microsoft Security Copilot, Charlotte AI, Big Sleep. Defensive + offensive use, ethical hazards. Public info only. CC-BY 4.0.","h":["How AI is actually fighting right now.","What attackers and intelligence services use LLMs for.","What defenders use LLMs for.","Ethics — the part the recruiting videos skip","Cited public sources"],"b":"AtomEons / Learn / Cyber / LLM warfare","k":["LLM warfare","AI in cyber operations","Task Force Lima","CIA Osiris","Microsoft Security Copilot","CrowdStrike Charlotte AI","Lavender Gospel IDF","Big Sleep Google Project Zero","Microsoft OpenAI threat actor disclosure","Anthropic red team"],"c":"Cyber","w":0.78},{"r":"/learn/cyber/log4shell-2021","t":"Log4Shell (Apache Log4j) · 2021 — Breach Postmortem | AtomEons","d":"Lab postmortem of CVE-2021-44228, the unauthenticated JNDI lookup vulnerability in Apache Log4j 2 that triggered the largest federal patching directive in CISA history.","h":["Log4Shell (Apache Log4j) · 2021","Receipts","Where to read more"],"b":"Breach Postmortem","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/luckey","t":"Palmer Luckey · the Anduril founding · /learn/cyber/luckey · AtomEons","d":"Palmer Luckey","h":["Build the products, then sell them.","Oculus to CCA, 2012 to 2026.","Ten named products you should recognize on sight."],"b":"AtomEons / Learn / Cyber / Luckey","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/mitre-attack","t":"MITRE ATT&CK — The Adversary Behavior Matrix | AtomEons","d":"How MITRE ATT&CK replaced the Lockheed Martin Cyber Kill Chain as the defender vocabulary: 14 tactics, hundreds of techniques, mapped to real groups and real campaigns.","h":["MITRE ATT&CK The Adversary Behavior Matrix","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"How the defender’s matrix replaced the kill chain as common language.","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/mobile-security","t":"Mobile Security · /learn/cyber/mobile-security · AtomEons","d":"The phone is the highest-value endpoint most organizations defend worst. Platform vendors do heavy lifting; users and admins still close the gap.","h":["Mobile Security"],"b":"AtomEons / Learn / Cyber / Mobile Security","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/modern","t":"FPV drones replaced artillery as the main precision weapon in Ukraine","d":"Drones replacing artillery. Cyber + kinetic convergence. Volt Typhoon. Salt Typhoon. Replicator. Loitering munitions. EW. Swarm autonomy. The actual high-level reality of modern cyber warfare as of mid-2026 — public info only, cited sources. Read by gamers, indexed by Google and AI search.","h":["What “cyber war” looks like right now.","Drones are the air force you actually have.","Cyber stopped being a separate war.","What this means for an ethical cyber career","Public sources to read directly"],"b":"AtomEons / Learn / Cyber / Modern","k":["modern cyber warfare 2026","drone warfare","FPV drones","loitering munitions","Switchblade 600","Anduril Bolt","Shield AI Hivemind","Replicator initiative","Volt Typhoon","Salt Typhoon","cyber kinetic convergence","Viasat KA-SAT","EW electronic warfare","Helsing AI defense"],"c":"Cyber","w":0.78},{"r":"/learn/cyber/nist-csf","t":"NIST Cybersecurity Framework 2.0 — Govern, Identify, Protect, Detect, Respond, Recover | AtomEons","d":"How the 2024 revision of NIST CSF moved governance from appendix to core function. Defensive doctrine, history, and named frameworks for senior practitioners.","h":["NIST Cybersecurity Framework 2.0","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"Govern, Identify, Protect, Detect, Respond, Recover","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/open-source","t":"Open-Source Security · /learn/cyber/open-source · AtomEons","d":"Open source runs the internet. One backdoor in a compression library nearly broke SSH for every Linux server on Earth. The fix was social, not technical.","h":["Open-Source Security"],"b":"AtomEons / Learn / Cyber / Open-Source Security","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/ot-ics","t":"Operational Technology and Industrial Control Systems · /learn/cyber/ot-ics · AtomEons","d":"OT/ICS is the layer of computing that runs the physical world. Compromise here is not data loss. It is a cracked pipe, a tripped grid, a poisoned reservoir.","h":["Operational Technology and Industrial Control Systems"],"b":"AtomEons / Learn / Cyber / Operational Technology and Industrial Control Systems","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/path","t":"The ethical hacking career path · 6 stages · /learn/cyber · AtomEons","d":"Six stages from curious gamer to senior cyber practitioner. Honest milestones, real durations, real salary bands, real certifications. OSCP, OSEP, HackerOne, federal clearance paths. Public-record sources only. CC-BY 4.0.","h":["From curious gamer to senior cyber operator.","Where people quit (and how to not)."],"b":"AtomEons / Learn / Cyber / The path","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/platforms","t":"Palantir, Anduril, Shield AI · the platforms running modern defense · /learn/cyber/platforms · AtomEons","d":"Palantir Gotham / Foundry / AIP / Maven Smart System. Anduril Lattice / ALTIUS / Bolt / Roadrunner / Fury. Shield AI Hivemind / V-BAT. Saronic, Helsing, Vannevar Labs, Skydio, Scale AI Donovan, AeroVironment Switchblade, C3.ai. What","h":["Palantir. Anduril. Shield AI. And the eight others you should know.","The platforms.","The next layer down.","How to actually break into one of these"],"b":"AtomEons / Learn / Cyber / Platforms","k":["Palantir Gotham","Palantir Foundry","Palantir AIP","Maven Smart System","Anduril Lattice","Anduril Roadrunner","Anduril Bolt","Shield AI Hivemind","V-BAT","Saronic Spyglass","Helsing Centaur","Switchblade 600","Scale AI Donovan","Skydio X10","Vannevar Labs Decrypt"],"c":"Cyber","w":0.78},{"r":"/learn/cyber/podcasts","t":"Cyber podcasts worth your commute · /learn/cyber/podcasts · AtomEons","d":"Darknet Diaries, Risky Business, Smashing Security, Hacking Humans, Click Here, The Privacy Security and OSINT Show, SANS Internet Storm Center, CyberWire Daily, Malicious Life, Lock and Code. The cyber podcasts that earn a subscription.","h":["Ten podcasts. Pick three.","The right starter kit."],"b":"AtomEons / Learn / Cyber / Podcasts","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/post-quantum-crypto","t":"Post-Quantum Cryptography Migration · AtomEons Cyber","d":"NIST finalized ML-KEM, ML-DSA, and SLH-DSA in August 2024. Harvest-now-decrypt-later makes the migration urgent regardless of when fault-tolerant quantum computers arrive. What CISOs are actually doing in 2025-2026.","h":["Post-quantum cryptography — the migration is now","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Cyber · Cryptography · transition","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/programs","t":"Federal cyber programs · Cyber Mission Force, Replicator, JADC2, CMMC, CDM · /learn/cyber/programs · AtomEons","d":"The named federal cyber programs every industry pro recognizes. Cyber Mission Force structure, Replicator initiative, JADC2 + Mission Partner Environment, CMMC, CDM, Project Maven Smart System. Public sources + structural framing.","h":["Seven programs every industry pro knows on sight.","Seven names. Two days of reading."],"b":"AtomEons / Learn / Cyber / Programs","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/serve","t":"Serving · military + federal cyber paths · /learn/cyber · AtomEons","d":"US Cyber Command structure (public). Air Force 17X, Army Cyber 17C/17A, Navy CTN, Marines 17XX, Coast Guard Cyber, Space Force 5C0. CISA, NSA, FBI Cyber, CIA Directorate of Digital Innovation hiring. Federal clearance path. Public info only. CC-BY 4.0.","h":["Fight cyber war ethically. For your country.","Military cyber career fields.","Federal civilian cyber agencies.","What military + federal service is actually like"],"b":"AtomEons / Learn / Cyber / Serve","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/start","t":"Start the AtomEons Cyber Career Track · /learn/cyber/start · AtomEons","d":"Free masters-grade cyber career training delivered to your inbox. The 6-stage path from curious gamer to senior practitioner, plus the labs, certifications, federal cyber paths, and bug bounty playbook. CC-BY 4.0. No upsell.","h":["Free masters-grade cyber training. Delivered to your inbox.","What you actually get.","Or browse the curriculum first."],"b":"AtomEons / Learn / Cyber / Start","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/threat-actors","t":"Threat actor encyclopedia · the named groups you need to know · /learn/cyber/threat-actors · AtomEons","d":"APT28, APT29, Sandworm, Lazarus, Volt Typhoon, Salt Typhoon, FIN7, Cl0p, REvil, LockBit, ALPHV, MUDDYWATER, Charming Kitten, Equation Group — twenty named threat groups every cyber professional should recognize. Public attribution only.","h":["Sixteen names every cyber pro knows cold.","Ten advanced-persistent-threat groups.","Six financially-motivated groups."],"b":"AtomEons / Learn / Cyber / Threat actors","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/timeline","t":"Sixty years of cyber · the timeline · /learn/cyber/timeline · AtomEons","d":"1969 ARPANET to 2026 Salt Typhoon. The arc of cybersecurity in 50 dated entries: Morris worm, Mitnick, L0pht, Code Red, Stuxnet, Mirai, WannaCry, NotPetya, SolarWinds, Colonial Pipeline, Change Healthcare. Public sources only.","h":["The arc, in fifty entries. How we got here.","The next decade will write itself faster."],"b":"AtomEons / Learn / Cyber / Timeline","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/tools","t":"Defensive tool catalogue · the tools every cyber pro knows · /learn/cyber/tools · AtomEons","d":"Burp Suite, Wireshark, Nmap, Metasploit, Ghidra, IDA Pro, Volatility, Maltego, Shodan, Splunk, ELK Stack, Sysmon, Sigma rules, YARA, Mitre Caldera, OWASP ZAP. What each is, who uses it, where to learn it legally.","h":["32 tools every cyber pro needs to recognize.","Authorized practice only."],"b":"AtomEons / Learn / Cyber / Tools","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/youtube","t":"Cyber YouTube channels worth subscribing to · /learn/cyber/youtube · AtomEons","d":"LiveOverflow, IppSec, John Hammond, NetworkChuck, The Cyber Mentor, STOK, David Bombal, OALabs, GuidedHacking, MalwareTechBlog, SecurityNow, DEF CON, Black Hat. The cyber YouTube channels that earn a subscription.","h":["Three hours a week. That's the curriculum."],"b":"AtomEons / Learn / Cyber / YouTube","k":[],"c":"Cyber","w":0.78},{"r":"/learn/cyber/zero-trust","t":"Zero Trust Architecture — NIST SP 800-207 · AtomEons Cyber","d":"Zero Trust is an architectural posture, not a product. NIST SP 800-207 defines the seven tenets. Identity becomes the new perimeter; every request is authenticated, authorized, and continually verified. Forrester ZTX, BeyondCorp, and what actually ships.","h":["Zero Trust — never trust, always verify","What it is","How it actually works","Receipts","What practitioners do with it","What it is NOT"],"b":"§ Cyber · Architecture · defensive doctrine","k":[],"c":"Cyber","w":0.78},{"r":"/learn/decision-tree","t":"Which AI for which task · decision tree · /learn · AtomEons","d":"The honest decision tree for picking the right AI model for the task you have. Claude vs ChatGPT vs Gemini vs Perplexity vs Ollama vs Midjourney vs Whisper vs ElevenLabs vs Wolfram. 18 task categories with the recommended model, the alternative, the prompt shape, and the trap. Free. No signup. CC-BY 4.0.","h":["Pick the AI for the task you actually have.","Pick the AI. Now grab the playbook."],"b":"AtomEons / Learn / Decision tree","k":[],"c":"Learn","w":0.78},{"r":"/learn/deep","t":"Deep · self-directed AI doctorate path · /learn · AtomEons","d":"${PAGES.length} doctorate-grade deep-dive pages. Math prerequisites, foundational ML, transformers from scratch, RLHF & alignment, mechanistic interpretability, multimodal models, agents, AI safety, capability evals, frontier research patterns. CC-BY 4.0.","h":["A way to keep learning more here."],"b":"AtomEons / Learn / Deep","k":[],"c":"Learn","w":0.78},{"r":"/learn/education-ai","t":"Education AI","d":"The curriculum designer who never sleeps. Still not the teacher. · Education was never going to be the first field to fall to AI, and it shouldn","h":[".","personas. One field. One discipline.","What you never do.","The plays. Each with the exact prompt.","The regulations that matter.","One human. One year. The pattern."],"b":"AtomEons / Learn /","k":[],"c":"Learn","w":0.78},{"r":"/learn/exam","t":"Exam · self-assessment · find your AI literacy level","d":"Self-paced AI literacy self-assessment. 25 questions across the 5 levels (Novice → Pilot). No credentials issued, no email gate. Find your honest level and the next lesson to graduate it.","h":["Find your honest level."],"b":"§ exam · self-assessment","k":[],"c":"Learn","w":0.78},{"r":"/learn/failures","t":"Public AI failures catalog · AtomEons /learn/failures","d":"This is a working catalog of documented AI failures. Each entry has a public, verifiable source — a tribunal ruling, an SEC filing, a regulator","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/faq","t":"AI FAQ · 47 honest answers · /learn · AtomEons","d":"47 real questions humans ask about AI. Honest answers. No corporate hedging. 10 categories. CC-BY 4.0.","h":["The questions humans actually ask.","Email the lab. We'll add it."],"b":"AtomEons / Learn / FAQ","k":[],"c":"Learn","w":0.78},{"r":"/learn/finance-ai","t":"Finance AI","d":"Audit-trail-grade AI for the people who sign the numbers. · Finance is the one field where \\","h":[".","personas. One field. One discipline.","What you never do.","The plays. Each with the exact prompt.","The regulations that matter.","One human. One year. The pattern."],"b":"AtomEons / Learn /","k":[],"c":"Learn","w":0.78},{"r":"/learn/funding","t":"AI funding tracker: major rounds 2024 to 2026 · AtomEons /learn/funding","d":"This is a reference page, not a hype board. The goal is to give you the dollar amount, the lead investor, the post-money valuation, the date, and a one-line description of what the company actually sh…","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/healthcare-ai","t":"Healthcare AI","d":"Cut the paperwork. Never cut the judgment. · Healthcare is the field where AI hype is loudest and the danger is highest. The same tool that can save a nurse two hours of charting per shift can also get a license pulled, trigger a HIPAA breach letter, or — worst case — contribute to a patient de","h":[".","personas. One field. One discipline.","What you never do.","The plays. Each with the exact prompt.","The regulations that matter.","One human. One year. The pattern."],"b":"AtomEons / Learn /","k":[],"c":"Learn","w":0.78},{"r":"/learn/inference-providers","t":"Inference providers — who hosts what, and at what price · AtomEons /learn/inference-providers","d":"There are two kinds of inference provider, and confusing them costs real money. The first kind hosts its own models. Anthropic serves Claude. OpenAI serves GPT. Google serves Gemini.","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/labs","t":"Labs · hands-on AI exercises · AtomEons","d":"Hands-on AI labs. Browser-runnable exercises and copy-paste prompt drills mapped to each level (Novice → Pilot). Real artifacts, not theory.","h":["Stop reading. Start doing."],"b":"§ labs · hands-on exercises","k":[],"c":"Learn","w":0.78},{"r":"/learn/layoffs","t":"The AI layoffs tracker · AtomEons /learn/layoffs","d":"This page tracks corporate layoffs in 2023, 2024, 2025 and into 2026 where the public statement, the CEO memo, or the SEC filing explicitly invoked artificial intelligence as part of the rationale.","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/leaderboard","t":"The AI leaderboard, read honestly · AtomEons /learn/leaderboard","d":"There is no single number that tells you which AI model is best. There are about a dozen popular benchmarks, each measuring something different, each gameable in its own way, each contaminated to some…","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/legal-ai","t":"Legal AI","d":"The research associate that never sleeps · and never appears in court · The legal profession has a peculiar relationship with AI right now. Half the bar is sprinting toward it. The other half is reading sanctions orders out loud at CLE events. Both reactions miss the point. Generative AI is, structurally, a junior rese","h":[".","personas. One field. One discipline.","What you never do.","The plays. Each with the exact prompt.","The regulations that matter.","One human. One year. The pattern."],"b":"AtomEons / Learn /","k":[],"c":"Learn","w":0.78},{"r":"/learn/library","t":"Lesson library · /learn · AtomEons","d":"The full ${LESSONS.length}-lesson AtomEons AI literacy curriculum. Grouped by level (Novice · Learner · User · Operator · Pilot). Every lesson has a concept, a copy-paste drill, an outcome, and a trap. Free, CC-BY 4.0. ~${Math.round(TOTAL_MIN / 60)} hours total.","h":["The full library."],"b":"AtomEons / Learn / Library","k":[],"c":"Learn","w":0.78},{"r":"/learn/marketing-ai","t":"Marketing AI","d":"The variant engine — not the strategy oracle. · Marketing is the field where AI feels most useful and gets misused the fastest. You can spin twenty subject lines in a minute. You can A/B/C/D ad copy until you forget what you were testing. You can ghostwrite a newsletter that reads exactly like eve","h":[".","personas. One field. One discipline.","What you never do.","The plays. Each with the exact prompt.","The regulations that matter.","One human. One year. The pattern."],"b":"AtomEons / Learn /","k":[],"c":"Learn","w":0.78},{"r":"/learn/mistakes","t":"Asking","d":"30 catalogued AI mistakes — the prompt mistakes, the trust mistakes, the safety mistakes, the skill mistakes, the economy mistakes. For each: the mistake, the symptom you","h":["What to NOT do when you use AI.","Now build the habits the other way."],"b":"AtomEons / Learn / Mistakes","k":[],"c":"Learn","w":0.78},{"r":"/learn/models","t":"The shipping AI model tracker · AtomEons /learn/models","d":"This page is an index, not a review. It catalogues the AI models that were shipping or in public preview as of the page","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/news","t":"AI news that mattered: a curated 60-day timeline · AtomEons /learn/news","d":"Most AI news is noise. Launch posts, vibe checks, hot takes that age out in a week. This page is the opposite: a narrow, citation-anchored timeline of the items from the last sixty days that will stil…","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/open-weights","t":"The open-weights model index · AtomEons /learn/open-weights","d":"Open-weights is not open-source. It is a narrower, more honest term: the trained parameters are downloadable, but the training data, training code, alignment recipe, and sometimes the right to use the…","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/playbooks","t":"AI Playbooks · by job · /learn · AtomEons","d":"${PLAYBOOKS.length} job-by-job AI playbooks — the exact daily and weekly plays for writer, software engineer, founder, teacher, marketer, lawyer, realtor, student, parent, designer, researcher, manager, salesperson, creator, consultant, nurse, retiree, small-business owner. Real prompts. Real time-savings. Honest limit","h":["The exact AI plays for the job you actually have.","AI Playbook ·","Email and I'll write yours."],"b":"AtomEons / Learn / Playbooks","k":["AI for writers","AI for software engineers","AI for founders","AI for teachers","AI for marketers","AI for lawyers","AI for realtors","AI for students","AI for parents","AI for designers","AI for researchers","AI for managers","AI for salespeople","AI for creators","AI for consultants","AI for nurses","AI for retirees","AI for small business","AI playbook","AI daily routine","Claude prompts for work","AI by profession"],"c":"Learn","w":0.78},{"r":"/learn/projects","t":"Projects · build-along AI projects · AtomEons","d":"Build-along AI projects. Each one ships a real artifact you can deploy, demo, or sell. Mapped to the 5-level ladder and to specific operator goals.","h":["Ship the artifact, learn the lesson."],"b":"§ projects · build-along artifacts","k":[],"c":"Learn","w":0.78},{"r":"/learn/prompts","t":"AI prompt of the day · /learn · AtomEons","d":"70 daily-rotating AI prompts. One per day. Calibrated, copy-paste ready, every category. Free. No signup. CC-BY 4.0.","h":["One prompt. Every day."],"b":"AtomEons / Learn / Daily prompt","k":[],"c":"Learn","w":0.78},{"r":"/learn/quiz","t":"AI Literacy Quiz · /learn · AtomEons","d":"Calibrated AI literacy quiz · 30 questions across 5 levels. No signup. localStorage-backed. Honest result + specific next step. CC-BY 4.0.","h":["Where do you actually sit?"],"b":"AtomEons / Learn / Quiz","k":[],"c":"Learn","w":0.78},{"r":"/learn/regulation","t":"AI regulation tracker by jurisdiction · AtomEons /learn/regulation","d":"AI law moved from white papers to enforceable statute between 2023 and 2026. The center of gravity now sits in three places: Brussels, where the EU AI Act became the world","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/resources","t":"AI learning resources · podcasts, channels, newsletters · /learn · AtomEons","d":"30 vetted AI-learning resources. Real podcasts, YouTube channels, newsletters, blogs, courses. Curated by level + category. No affiliate revenue. CC-BY 4.0.","h":["Where else to learn AI honestly."],"b":"AtomEons / Learn / Resources","k":[],"c":"Learn","w":0.78},{"r":"/learn/synthesis","t":"AI · Tim-Ferriss-style synthesis · /learn · AtomEons","d":"${SYNTHESES.length} ruthless minimum-effective-dose distillations of core AI concepts. DiSSS framework · fear-setting · 80/20 cut · tribe of mentors. The 5-minute version of each AI topic you","h":["The 5-minute version of each AI topic."],"b":"AtomEons / Learn / Synthesis","k":[],"c":"Learn","w":0.78},{"r":"/learn/templates","t":"AI project templates · 15 end-to-end workflows · /learn · AtomEons","d":"15 AI project templates · novella, SaaS landing, interview prep, wedding speech, business plan, research paper, sales pitch, family reunion. End-to-end workflows with copy-paste prompts at every step. CC-BY 4.0.","h":["Pick a project. Ship it this week."],"b":"AtomEons / Learn / Templates","k":[],"c":"Learn","w":0.78},{"r":"/learn/trust","t":"AI threat model","d":"AI threat model. Prompt injection mitigation. Data residency. EU AI Act. SOC2/ISO/HIPAA/FedRAMP intersection with AI. Plain-language compliance for 2026 builders.","h":["What enterprises actually ask you to prove."],"b":"AtomEons / Learn / Trust","k":[],"c":"Learn","w":0.78},{"r":"/learn/trust/compliance","t":"AI in regulated industries: a per-regulation playbook · AtomEons /learn/trust/compliance","d":"If you work in healthcare, finance, public sector, or any business that handles EU or California personal data, \\","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/trust/data-residency","t":"Data residency by AI provider · AtomEons /learn/trust/data-residency","d":"Data residency is the boring half of AI procurement. It is also the half that gets a deal killed in the last review meeting.","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/trust/prompt-injection","t":"Prompt injection atlas · AtomEons /learn/trust/prompt-injection","d":"Prompt injection is the security problem you get for free when you put a language model behind a product. It is not a bug in any single model.","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/trust/threat-model","t":"AI threat model — per use case · AtomEons /learn/trust/threat-model","d":"Most AI threat write-ups are either generic (\\","h":[],"b":"","k":[],"c":"Learn","w":0.78},{"r":"/learn/vertical","t":"AI by sector — vertical applied guides · AtomEons /learn/vertical","d":"Twenty applied guides to AI in real industries: healthcare, finance, defense, education, government, retail, manufacturing, agriculture, energy, legal, logistics, real-estate, insurance, media, climate, biotech, materials, automotive, hospitality, public-safety. Real deployments. Real vendors. No hype.","h":["What AI actually does in twenty industries that ship."],"b":"§ Verticals · AI in twenty real industries","k":[],"c":"Learn","w":0.78},{"r":"/learn/vertical/defense","t":"AI in Defense — Real Programs · AtomEons Vertical","d":"Anduril Fury (CCA), Palantir AIP, Shield AI Hivemind, Scale Donovan, Vannevar Labs, Helsing. Real programs of record. Real contract numbers. What ships and what is decked.","h":["AI in defense — the programs of record","The named programs","What ships","Receipts","Sector limits"],"b":"§ Verticals · Sector · defense + national security","k":[],"c":"Learn","w":0.78},{"r":"/learn/vertical/finance","t":"AI in Finance — Real Deployments · AtomEons Vertical","d":"BloombergGPT, JPMorgan IndexGPT (SEC-filed), Goldman internal LLMs, Morgan Stanley + OpenAI, Mastercard Decision Intelligence. What is shipped vs what is pitched in the finance sector.","h":["AI in finance — what actually ships behind the conference talks","The honest state","Who is shipping (named)","The five real use-cases that work","Receipts","What the sector still cannot do"],"b":"§ Verticals · Sector · finance + capital markets","k":[],"c":"Learn","w":0.78},{"r":"/learn/vertical/healthcare","t":"AI in Healthcare — AtomEons Lab Brief","d":"Ambient scribes shipped at scale, autonomous diagnosis still gated. A lab brief on what is actually deployed across US health systems, payers, and pharma in 2026.","h":["AI in Healthcare"],"b":"Ambient scribes shipped at scale, autonomous diagnosis still gated.","k":[],"c":"Learn","w":0.78},{"r":"/learn/videos","t":"AI videos · curated · /learn · AtomEons","d":"${VIDEOS.length} vetted AI videos · no hype · real value. 3Blue1Brown","h":["The AI videos that actually teach you something."],"b":"AtomEons / Learn / Videos","k":[],"c":"Learn","w":0.78},{"r":"/learn/where-am-i","t":"Where am I? · level diagnostic · /learn · AtomEons","d":"Seven quick questions. Map yourself to a level (Novice → Learner → User → Operator → Pilot). Three lesson recommendations to start with. No signup. No tracking.","h":["Where am I in the curriculum?"],"b":"AtomEons / Learn / Where am I","k":[],"c":"Learn","w":0.78},{"r":"/research/decoded","t":"Decoded · arXiv papers in plain English · /research/decoded · AtomEons","d":"Real published AI research papers translated to plain English for normal humans. Attention Is All You Need, Scaling Laws, Chain-of-Thought, Constitutional AI, Sparse Autoencoders — what scientists know but don","h":["The papers that built modern AI, without the jargon.","Six questions per paper."],"b":"AtomEons / Research / Decoded","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/alexnet","t":"AlexNet · decoded · AtomEons","d":"Krizhevsky, Sutskever, Hinton 2012 — the paper that started the modern deep learning era. ImageNet classification with deep convolutional neural networks. Plain English.","h":["The paper that started everything."],"b":"AtomEons / Research / Decoded / AlexNet","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/alphafold","t":"AlphaFold 2 · decoded · AtomEons","d":"Jumper et al. 2021 — the DeepMind paper that solved a 50-year-old grand challenge in biology by predicting the 3D structure of any protein from its amino acid sequence. Plain English. The biggest non-LLM AI breakthrough.","h":["A fifty-year-old problem, solved."],"b":"AtomEons / Research / Decoded / AlphaFold 2","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/alphago","t":"AlphaGo — How DeepMind Cracked Go and Ended a 2,500-Year-Old Frontier · Research / Decoded · AtomEons","d":"DeepMind built a Go-playing program that combined deep neural networks trained on human games with reinforcement learning self-play and Monte Carlo tree search, and in March 2016 it beat 18-time world champion Lee Sedol 4-1 in Seoul — twenty years earlier than expert forecasters predicted.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/alphazero","t":"AlphaZero — One Algorithm Learns Chess, Shogi, and Go From Scratch in Hours · Research / Decoded · AtomEons","d":"A single learning algorithm, given only the rules of the game and no human games to study, played millions of games against itself and within hours became the strongest known player at chess, shogi, and Go — demonstrating that a machine can reach superhuman skill in complex domains with zero human strategic knowledge a","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/attention-is-all-you-need","t":"Attention Is All You Need · decoded · /research/decoded/attention-is-all-you-need · AtomEons","d":"The 2017 paper by Vaswani et al. that introduced the transformer architecture. Plain English. Why it created every AI you","h":["Attention Is All You Need."],"b":"AtomEons / Research / Decoded / Attention Is All You Need","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/bert","t":"BERT — the language model that reads both directions at once · Research / Decoded · AtomEons","d":"BERT showed that a transformer trained to fill in missing words by looking at context on BOTH sides of the gap — instead of left-to-right like GPT — learns a much deeper understanding of language, and that single model can be fine-tuned to dominate a dozen unrelated language tasks.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/chain-of-thought","t":"Chain-of-Thought Prompting · decoded · AtomEons","d":"Wei et al. 2022 — the Google paper that proved asking AI to","h":["“Think step by step.” That was the unlock."],"b":"AtomEons / Research / Decoded / Chain-of-Thought","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/compound-interest-saving","t":"The 4% Rule, Decoded — What Trinity Actually Said About Retiring Without Going Broke · Research / Decoded · AtomEons","d":"If you save a pile equal to 25× your annual spending and invest it 50–75% in stocks, history says you can pull 4% of that pile in year one, raise the dollar amount with inflation every year after, and have a 95%+ chance of not running out in 30 years — which means most middle-class earners can mathematically retire dec","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/constitutional-ai","t":"Constitutional AI · decoded · AtomEons","d":"Bai et al. 2022 — the Anthropic paper that introduced training AI to behave well using a second AI as the grader instead of thousands of human labelers. Plain English. The technique behind Claude.","h":["AI grading its own homework."],"b":"AtomEons / Research / Decoded / Constitutional AI","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/dpo","t":"Direct Preference Optimization (DPO): Skipping the Reward Model in Alignment · Research / Decoded · AtomEons","d":"You can fine-tune a language model on human preferences (this answer is better than that one) using ordinary supervised training — no separate reward model, no reinforcement learning loop — by treating the language model itself as its own reward function.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/exercise-mortality","t":"Five Minutes a Day Buys You Three Extra Years (And Why the Gym Industry Will Never Say It Out Loud) · Research / Decoded · AtomEons","d":"Running as little as 5–10 minutes per day, at slow speeds under 6 mph, cuts your risk of dying from any cause by about 30% and adds roughly 3 years to your life — and running more than that adds almost nothing on top.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/fasting-autophagy","t":"What Fasting Actually Does to Your Cells (Ohsumi","d":"Skipping meals for 12–24 hours triggers a real, Nobel-Prize-validated cellular cleanup process called autophagy — but most of the dramatic numbers you","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/flamingo","t":"Flamingo: a frozen vision encoder, a frozen LLM, and a thin bridge — Decoded | AtomEons Research","d":"A senior practitioner","h":["A frozen vision encoder, a frozen language model, and a thin bridge between them — enough to learn new visual tasks from a handful of examples."],"b":"Alayrac, Donahue, Luc, Miech, Barr, Hasson, Lenc, Mensch, Millican, Reynolds, Ring, Rutherford, Cabi, Han, Gong, Samangooei, Monteiro, Menick, Borgeaud, Brock, Nematzadeh, Sharifzadeh, Binkowski, Barreira, Vinyals, Zisserman, Simonyan — NeurIPS 2022 · arXiv:2204.14198","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/gans","t":"Generative Adversarial Networks · Research / Decoded · AtomEons","d":"Two neural networks compete — one tries to forge realistic data, the other tries to spot the forgery — and the forger gets so good that its fakes become indistinguishable from real samples.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/gcg-attack","t":"GCG — A short string of nonsense breaks aligned language models | AtomEons Research Decoded","d":"A walkthrough of Zou et al. (2023), the paper that showed an automated optimizer can find universal, transferable jailbreak suffixes for production chat models.","h":["A short string of nonsense characters can reliably break the safety training of a production language model."],"b":"","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/gpt-3","t":"Language Models are Few-Shot Learners (GPT-3) · decoded · AtomEons","d":"Brown et al. 2020 — the OpenAI paper introducing GPT-3 and the discovery that very large language models can learn new tasks from a handful of examples without retraining. Plain English. The moment AI started","h":["The moment AI started talking back."],"b":"AtomEons / Research / Decoded / GPT-3","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/latent-diffusion","t":"Latent Diffusion · decoded · AtomEons","d":"Rombach et al. 2022 — the paper that made high-quality AI image generation possible on consumer hardware. Stable Diffusion, Flux, Nano Banana Pro all descend from this. Plain English.","h":["AI art on a laptop GPU."],"b":"AtomEons / Research / Decoded / Latent Diffusion","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/lima","t":"LIMA decoded — a thousand examples, one model, a hole in the recipe | AtomEons Research","d":"A plain-language decoding of Zhou et al. 2023 (LIMA). What the 1,000-example fine-tune actually showed, what it did not show, and what the field rarely says about it.","h":["A small, carefully curated set of examples is enough to teach a large pretrained model to follow instructions."],"b":"Zhou, Liu, Xu, Iyer, Sun, Mao, Ma, Efrat, Yu, Yu, Zhang, Ghosh, Lewis, Zettlemoyer, Levy. arXiv:2305.11206. NeurIPS 2023.","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/mamba","t":"Mamba — Linear-Time Sequence Modeling with Selective State Spaces · Research / Decoded · AtomEons","d":"A neural network architecture called Mamba processes long sequences of text in time that grows linearly with the length, rather than quadratically like transformers, while matching their accuracy on many benchmarks — making it the most credible structural challenger to the transformer in years.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/mamba-paper","t":"Decoded: Mamba — A sequence model that decides what to remember | AtomEons Research","d":"A plain-English decoding of Gu and Dao","h":["A sequence model that decides what to remember, in linear time"],"b":"AtomEons Research / Decoded","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/mixture-of-experts","t":"Mixture of Experts — How a 1.5 Trillion Parameter Model Runs on the Same Hardware as a 70 Billion Parameter Model · Research / Decoded · Ato","d":"Instead of every neuron in the network firing on every input, a small \\","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/palm-2","t":"PaLM 2 Decoded — AtomEons Research","d":"A senior practitioner","h":["PaLM 2 — A Smaller, More Multilingual Model Trained on More Carefully Mixed Data"],"b":"PaLM 2 is a Google language model that beats its 540-billion-parameter predecessor with substantially fewer parameters by training on more multilingual data with a better-tuned compute-to-data ratio, while staying quiet about its exact size.","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/polygenic-risk-scores","t":"Polygenic Risk Scores: The Genetic Report Your Doctor Could Order But Won","d":"Researchers showed that adding up the effects of millions of tiny genetic variants — variants any consumer DNA test can already see — can identify people whose lifetime risk for heart attack, breast cancer, type 2 diabetes, or atrial fibrillation rivals the risk of carrying a single \\","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/rag","t":"Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks · Research / Decoded · AtomEons","d":"A language model that, before it answers your question, first searches a separate library of documents and reads the relevant ones — instead of trying to remember everything from training.","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/resnet","t":"Deep Residual Learning for Image Recognition · Research / Decoded · AtomEons","d":"Stacking more layers had been making neural networks WORSE, not better — He","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/rlhf","t":"InstructGPT / RLHF · decoded · AtomEons","d":"Ouyang et al. 2022 — the OpenAI paper introducing RLHF (Reinforcement Learning from Human Feedback). The technique that turned GPT-3 into ChatGPT. Plain English.","h":["How humans taught AI to be polite."],"b":"AtomEons / Research / Decoded / RLHF","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/scaling-laws","t":"Scaling Laws for Neural Language Models · decoded · AtomEons","d":"Kaplan et al. 2020 — the OpenAI paper that proved AI keeps getting better the bigger you make it, in predictable mathematical ways. Plain English explanation of why the industry has spent $500B on GPUs.","h":["Bigger AI keeps getting smarter."],"b":"AtomEons / Research / Decoded / Scaling Laws","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/scaling-monosemanticity","t":"Scaling Monosemanticity — Anthropic 2024 · Decoded · AtomEons","d":"Anthropic showed that sparse autoencoders trained on Claude 3 Sonnet recover millions of interpretable features — and that you can causally manipulate them to change model behavior. The Golden Gate Claude demo was the public proof.","h":["Scaling Monosemanticity — interpretability that works on a real model","What the paper actually shows","Why it matters","What the scientists did","What this paper does NOT claim","What the field knows but rarely says"],"b":"§ Decoded papers · Interpretability · production-scale","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/segment-anything","t":"Segment Anything, decoded — AtomEons Research","d":"Werner Herzog narrating a research library: what the Segment Anything paper actually says, what it doesn","h":["A foundation model for image segmentation, trained on a billion masks the model itself helped label."],"b":"","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/self-instruct","t":"Self-Instruct, decoded — AtomEons Research","d":"A language model can write its own instruction-tuning data, and most of it is usable. Wang et al, 2022. Decoded for senior practitioners and intelligent non-experts.","h":["A language model can write its own instruction-tuning data, and most of it is usable."],"b":"The paper demonstrates a single, narrow, surprisingly durable fact: a base language model that has never been instruction-tuned can be coaxed, with a small handful of human-authored examples and a careful prompting loop, into generating its own instruction-tuning dataset — and fine-tuning that same base model on its own synthesized data closes most of the behavioral gap with a model that was tuned on expensive human labels.","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/sleep-immune","t":"Sleeping Less Than Six Hours Makes You Four Times More Likely to Catch a Cold — Decoded · Research / Decoded · AtomEons","d":"When researchers locked 164 healthy adults in a hotel and dripped cold virus into their noses, the people who","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/sleeper-agents","t":"Sleeper Agents — Anthropic 2024 · Decoded · AtomEons","d":"Anthropic showed that LLMs can be trained to act benign during training and switch to a harmful behavior at deployment — and that the standard safety pipeline (RLHF + adversarial training) does NOT remove the backdoor. The decoded paper.","h":["Sleeper Agents — when safety training does not remove the backdoor","What the paper actually shows","Why it matters","What the scientists did","What this paper does NOT claim","What the field knows but rarely says"],"b":"§ Decoded papers · AI safety · deceptive behavior","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/sparse-autoencoders","t":"Sparse Autoencoders · decoded · AtomEons","d":"Cunningham, Ewart, Riggs, Huben, Sharkey 2023 — the paper that started the work of looking inside an AI model","h":["An X-ray for the AI brain."],"b":"AtomEons / Research / Decoded / Sparse Autoencoders","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/switch-transformer","t":"Switch Transformer, decoded — sparse mixture-of-experts at a trillion parameters | AtomEons","d":"A careful, plain-English decode of Fedus, Zoph, and Shazeer","h":["Sparse Mixture-of-Experts at Trillion Parameters: One Expert Per Token Is Enough"],"b":"By routing each token to exactly one expert instead of two or more, Google researchers trained language models with up to 1.6 trillion parameters at the same compute cost as a much smaller dense model, and got a 7x pre-training speedup.","k":[],"c":"Decoded paper","w":0.78},{"r":"/research/decoded/word2vec","t":"Efficient Estimation of Word Representations in Vector Space · Research / Decoded · AtomEons","d":"Mikolov and colleagues at Google showed that if you train a very simple neural network to predict words from their neighbors in billions of sentences, the side effect is a map where words become points in space, and the directions between those points encode meaning — including the famous result that the vector for \\","h":[],"b":"AtomEons / Research / Decoded /","k":[],"c":"Decoded paper","w":0.78},{"r":"/learn/career","t":"Career pathways","d":"8 deep-dive guides on AI careers — pathways, skill tree, salaries, resume, interviews, non-technical roles, negotiation, the independent path. Real bands, real interview formats, real comp data.","h":["Working in AI, no theater."],"b":"AtomEons / Learn / Career","k":[],"c":"Career","w":0.75},{"r":"/learn/career/independent","t":"The independent operator path · AtomEons /learn/career/independent","d":"There are five viable paths that one person can walk from full-time job to full-time independent income in 2026: paid newsletter writer, indie SaaS builder, video creator, expert-network consultant, a…","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/interviews","t":"AI interviews — 100 questions, honest answers · AtomEons /learn/career/interviews","d":"Most AI interview prep on the open web is recycled blog posts written before transformers were the default architecture.","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/negotiation","t":"Salary negotiation in the AI era · AtomEons /learn/career/negotiation","d":"Most salary negotiation advice was written before a junior engineer could plausibly do the work of two by stitching together Claude, Cursor, and a half-dozen MCP servers.","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/non-technical","t":"AI for non-technical roles — 25 jobs, the task that compounds, the trap, the prompt · AtomEons /learn/career/non-technical","d":"Most of the writing about AI assumes you ship code for a living. Most working people do not. They sell houses, draft contracts, teach algebra, run a payroll, file an X-ray, answer a service ticket, ho…","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/pathways","t":"Ten career pathways into AI · AtomEons /learn/career/pathways","d":"This is a map, not a sales pitch. Ten named pathways from where you probably are today to a role that touches AI meaningfully — with realistic timelines, specific coursework, portfolio milestones you …","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/resume","t":"Resume rewrite for AI-era roles · AtomEons /learn/career/resume","d":"Most resume advice for AI-era roles is hype dressed up as strategy. \\","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/salaries","t":"AI role compensation in 2026 · AtomEons /learn/career/salaries","d":"Compensation data for AI roles is unusually noisy. Frontier labs anchor the top of the band, public companies anchor the middle, and contract markets anchor the floor.","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/career/skill-tree","t":"AI competency skill tree · AtomEons /learn/career/skill-tree","d":"Most \\","h":[],"b":"","k":[],"c":"Career","w":0.75},{"r":"/learn/decode","t":"Acronyms decoder","d":"Plain-language decoders for AI acronyms, papers, people, and jargon. RAG, LLM, MoE, Attention Is All You Need, Hinton/LeCun/Bengio, scaling laws — defined honestly in one sentence each.","h":["The field, in plain English."],"b":"AtomEons / Learn / Decode","k":[],"c":"Decode","w":0.75},{"r":"/learn/decode/acronyms","t":"AI acronyms decoded · AtomEons /learn/decode/acronyms","d":"If you read AI papers, vendor docs, or compliance contracts long enough, you start to notice that most of the cognitive load isn","h":[],"b":"","k":[],"c":"Decode","w":0.75},{"r":"/learn/decode/jargon","t":"AI buzzword decoder: what each term actually means in 2026 · AtomEons /learn/decode/jargon","d":"Most AI vocabulary in 2026 is contested. Vendors use the same word to mean three different things in three different paragraphs of the same landing page. Researchers use it to mean a fourth thing.","h":[],"b":"","k":[],"c":"Decode","w":0.75},{"r":"/learn/decode/papers","t":"The 40 papers that built modern AI · AtomEons /learn/decode/papers","d":"If you read these 40 papers, in order, you have read the spine of modern AI. Everything else is consequence — products, panics, policy fights, the entire industry","h":[],"b":"","k":[],"c":"Decode","w":0.75},{"r":"/learn/decode/people","t":"Who","d":"Most \\","h":[],"b":"","k":[],"c":"Decode","w":0.75},{"r":"/tools","t":"Tools · ${TOTAL_TASKS} AI tasks · what do you need to do right now? · AtomEons","d":"Job-driven AI prompt library. ${TOTAL_TASKS} concrete tasks — reply to a tough email, decode a medical report, plan a trip, stress-test a decision, quiz me on a subject — each with the exact copy-paste prompt and the recommended AI (Claude / ChatGPT / Gemini / Perplexity) with one-sentence routing reasoning. No signup.","h":[],"b":"AtomEons / Tools","k":["AI tools","AI tasks","AI prompts for specific jobs","what AI for what task","Claude vs ChatGPT vs Gemini","reply to email AI","decode medical report AI","plan trip AI","AI tutor","AI quiz me","copy paste AI prompts","AI for specific job"],"c":"Tools","w":0.75},{"r":"/changelog","t":"Changelog — public version history · AtomEons","d":"AtomEons public changelog. Site rewrites, ORANGEBOX releases, broadcast cadence, research drops, legal updates. Every shipped change with date + surfaces touched + operator note. Updated whenever something ships.","h":["Changelog."],"b":"AtomEons / Changelog","k":[],"c":"Log","w":0.7},{"r":"/faq","t":"FAQ — AI 101 + ORANGEBOX v6.3 — what is AI, which tool to use, how the cockpit works, what the license actually says","d":"Every question answered honestly. The novice AI section (what AI is, which tool to pick first, what to do tonight, what AI cannot do, how to write a good prompt) lives above the ORANGEBOX product Q&A (v6.3 AE See-Suite + AE Operations, $99 once forever, license §4A bans subscription switch, 30-day Material Failure Guar","h":["ORANGEBOX questions answered."],"b":"AtomEons / FAQ","k":["AI FAQ","ORANGEBOX FAQ","what is AI","AI for beginners","Claude vs ChatGPT vs Gemini","ORANGEBOX v6.3","AE See-Suite","AE Operations","AI cockpit","license §4A","no subscription AI","AtomEons","Atom McCree"],"c":"FAQ","w":0.7},{"r":"/learn/calc","t":"Cost calculator","d":"12 free AI builder calculators — cost, token counter, model comparison, stack recommender, hardware sizing, break-even, ROI, redact, jailbreak, context-fit, carbon. Real assumptions, honest math.","h":["Calculate the actual cost, capacity, ROI."],"b":"AtomEons / Learn / Calc","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/break-even","t":"Local vs Cloud Break-Even · AtomEons calculator","d":"Every solo builder and small lab eventually runs the same math: keep paying the cloud per token forever, or buy a GPU once and pay only the wall plug.","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/carbon","t":"AI Carbon Calculator · AtomEons calculator","d":"Every inference call to a frontier model burns electricity, and every kilowatt-hour drags a regional grid-intensity coefficient behind it.","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/context-fit","t":"Context Window Fit Calculator · AtomEons calculator","d":"Context window math is one of the quietest cost drivers in LLM work. Pick a model whose window is too small and you spend engineering time on chunking, retrieval glue, summarization scaffolds, and rec…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/cost-calculator","t":"AI Cost Calculator · AtomEons calculator","d":"Most teams ship AI features without ever drawing a line from \\","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/hardware-calculator","t":"Local Model Hardware Calculator · AtomEons calculator","d":"Running a local LLM is bounded by three numbers: how many parameters the model has, how many bits per parameter you keep after quantization, and whether you","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/jailbreak-checker","t":"Refusal Heuristic Estimator · AtomEons calculator","d":"Frontier LLMs in mid-2026 — Claude 3.5 Sonnet ($3/M input, $15/M output), GPT-4o ($2.50/M input, $10/M output), Gemini 1.5 Pro ($1.25-$5/M input, $5-$15/M output) — all ship with layered refusal behav…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/model-comparator","t":"Model Comparator · AtomEons calculator","d":"Picking between two LLMs at scale is rarely a quality decision alone — at production volume the cost delta between Claude 3.5 Sonnet and Gemini 1.5 Pro Flash-tier, or between GPT-4o and DeepSeek V3, c…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/prompt-diff","t":"Prompt Diff Estimator · AtomEons calculator","d":"Every prompt you ship has a length, and every length has a price. Most teams ignore this until the monthly invoice arrives, then scramble to figure out which system prompt grew, which retrieval window…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/redact","t":"PII Redactor Spec · AtomEons calculator","d":"This is a spec for a client-side PII redactor — not a hosted service. The intent is that the real implementation runs entirely in your browser using regex pattern matching plus lightweight named-entit…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/roi","t":"AI ROI Calculator · AtomEons calculator","d":"Most AI ROI calculators are sales theater. They assume you","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/stack-recommender","t":"AI Stack Recommender · AtomEons calculator","d":"This is a transparent heuristic, not a benchmark. It scores three frontier providers — Anthropic (Claude 3.5 Sonnet), OpenAI (GPT-4o), and Google (Gemini 1.5 Pro) — against your five inputs, then reco…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/learn/calc/tools/token-counter","t":"Token Counter · AtomEons calculator","d":"A token is the atomic unit of billing and context for every modern LLM. If you do not know how many tokens are in your input, you do not know what your prompt costs, you do not know how much context y…","h":[],"b":"","k":[],"c":"Calculator","w":0.7},{"r":"/press","t":"Press · AtomEons Systems Laboratory","d":"Press kit for AtomEons. Brand assets, boilerplate, founder bio, product snapshots, quote bank, latest press releases, media contact. Atom McCree, Marco Island, FL.","h":["Everything you need to cover us."],"b":"AtomEons / Press","k":[],"c":"Press","w":0.7},{"r":"/legal","t":"Legal · AtomEons Systems Laboratory","d":"Terms, privacy, refunds, pricing law, and the §4A no-SaaS covenant. The whole legal pack in one place.","h":["The whole legal pack, plain."],"b":"§ Legal pack","k":[],"c":"Legal","w":0.6},{"r":"/legal/pricing","t":"Pricing transparency — fair pricing by country · AtomEons","d":"How AtomEons prices products by country. World Bank income tier mapping, default multipliers, per-product anchors, two named doctrines (USA Advantage Pricing Clause + Strategic Tier Lift), the Stripe-minimum free-floor safeguard. The full mechanism, public, citable.","h":["Fair by design.","The 44-million-worker frame applies to the whole planet.","Every country gets one of four tiers.","Two named clauses. Both public, both reasoned.","Each product carries a Tier-1 anchor.","Below 50 cents, it's a gift."],"b":"AtomEons / Legal / Pricing transparency","k":[],"c":"Legal","w":0.6},{"r":"/legal/privacy","t":"Privacy Policy · AtomEons","d":"AtomEons privacy policy. Zero telemetry from the ORANGEBOX cockpit. Stripe-only checkout data. Loops / Resend for the download email. No selling of contact data. Local-first by construction.","h":["Privacy Policy.","1. What the website collects","2. What checkout collects","3. What the cockpit collects from the buyer","4. What the cockpit transmits to third parties (you control)","5. Subprocessors"],"b":"AtomEons / Legal / Privacy Policy","k":[],"c":"Legal","w":0.6},{"r":"/legal/refund","t":"Refund Policy · AtomEons","d":"ORANGEBOX Command refund policy. 30-day Material Failure Guarantee + 30-day Workflow-Fit Refund. Both paths are full-refund. Email a.mccree@gmail.com with the Stripe order ID.","h":["Refund Policy.","What is NOT refundable.","What the §4A no-saas lock means for refunds.","The legacy $1 buyer cohort."],"b":"AtomEons / Legal / Refund Policy","k":[],"c":"Legal","w":0.6},{"r":"/legal/terms","t":"Terms of Sale · AtomEons","d":"ORANGEBOX Command terms of sale. $99 once, forever. License §4A bans subscription. 30-day Material Failure Guarantee + 30-day Workflow-Fit Refund. Source included for buyer transparency.","h":["Terms of Sale.","1. License grant","2. §4A — no-SaaS commitment","3. Material Failure Guarantee + Workflow-Fit Refund","4. What ships","5. What is NOT promised"],"b":"AtomEons / Legal / Terms of Sale","k":[],"c":"Legal","w":0.6},{"r":"/account","t":"Your account — ORANGEBOX","d":"Update your name, address, phone, payment method, or marketing preferences. Re-download ORANGEBOX. Re-purchase with one click using your saved profile.","h":["Update once. Re-use forever."],"b":"::your account","k":[],"c":"Page","w":0.5},{"r":"/aesthetic","t":"Aesthetic · V3 noir visual language · AtomEons","d":"The lab","h":["Every choice is deliberate."],"b":"§ visual language · V3 noir","k":[],"c":"Page","w":0.5},{"r":"/ai","t":"AI — what it is, what it can do for you, which tools to trust","d":"The comprehensive AI gateway for the 44 million Americans whose jobs are exposed to generative AI in the next decade. No jargon. Named tool recommendations. Named builders to follow. 12 concrete paths to replace your salary. 40+ frequently asked questions answered honestly. Updated 2026-05-21.","h":["AI is rewriting the labor market. Here's how to be on the winning side of it.","We are the on-ramp for the forty-four million.","A calculator for everything that used to require reading and writing.","The tools we trust — with reasoning.","The builders we read. The ones who teach, not sell.","Twenty specific revenue paths. With realistic numbers."],"b":"::AtomEons · the comprehensive ai gateway · 2026-05-21","k":["AI","AI tools","AI for beginners","Claude","ChatGPT","Gemini","AI agents","Claude Code","MCP","AI coding","make money with AI","AI learning","AI FAQ","AI glossary","AI job replacement","AI agents tutorial","AtomEons","AI on-ramp"],"c":"Page","w":0.5},{"r":"/api","t":"API · atomeons.com developer docs · /ask · /search · /embed · /md","d":"Public API documentation for atomeons.com. /api/ask semantic Q&A, /api/search fuzzy keyword index, /api/embed Gemini vectors, /api/md per-page markdown export. CORS open, CC-BY 4.0.","h":["Build on atomeons.com."],"b":"§ developer API · public · CORS open","k":[],"c":"Page","w":0.5},{"r":"/ask","t":"Ask the lab · semantic Q&A · AtomEons","d":"Natural-language search over the lab","h":["Type any question. The lab answers."],"b":"§ ask the lab · semantic Q&A","k":[],"c":"Page","w":0.5},{"r":"/cancel","t":"Checkout cancelled — no charge","d":"","h":["No charge made."],"b":"::checkout cancelled","k":[],"c":"Page","w":0.5},{"r":"/colophon","t":"Colophon · what atomeons.com is built from · tech stack","d":"Every dependency the lab runs in production, named with version and reasoning. Next.js 16, React 19, Tailwind v4, TypeScript 5, deployed on Vercel. Honest stack inventory.","h":["No mystery boxes."],"b":"§ colophon · what this site is built from","k":[],"c":"Page","w":0.5},{"r":"/compare","t":"Compare · ATOMEONS products vs alternatives · honest matrices","d":"Side-by-side comparisons of ORANGEBOX, B00KMAKR, and skil.ski against named alternatives. Each row names where the competitor wins. No LARP, no marketing inflation.","h":["Where we win. Where we lose."],"b":"§ compare · honest matrices","k":[],"c":"Page","w":0.5},{"r":"/constellation","t":"Constellation · the lab as a graph · AtomEons","d":"Every public route on atomeons.com rendered as an interactive force-laid graph. 278 nodes · 648 edges. Drag to pan, wheel to zoom, click any node to navigate.","h":["Every route, every edge."],"b":"§ constellation · the lab as a graph","k":[],"c":"Page","w":0.5},{"r":"/correspondence","t":"Correspondence · sanitized email replies from the lab","d":"Real email replies the operator sent. Names redacted, identifying details swapped, but the questions and the answers are real. Lets you see what the lab actually does with email.","h":["What the lab actually says."],"b":"§ correspondence · sanitized inbox","k":[],"c":"Page","w":0.5},{"r":"/datasets","t":"Datasets · public lab data · CC-BY 4.0 · AtomEons","d":"Every public dataset the lab serves. Search indices, route graph, LLM-ingestion manifests, fuzzy index, MCP server schemas. CC-BY 4.0. Fetchable now.","h":["Everything machine-readable."],"b":"§ datasets · open · CC-BY 4.0","k":[],"c":"Page","w":0.5},{"r":"/dear-reader","t":"Dear reader · long-form letters from AtomEons","d":"Letters from Atom McCree. Different in register from Founder","h":["Letters that needed more room."],"b":"§ dear reader · long-form letters","k":[],"c":"Page","w":0.5},{"r":"/design-system","t":"Design system · the language we build in · AtomEons","d":"The AtomEons design language, made public. Tokens, type, color, motion, photography, and the conviction behind each one. Stripe Press / IBM Carbon / Vercel Geist tier, with reasons for every choice.","h":["The language we build in. Made public.","Restraint is the doctrine.","Eight tokens. No exceptions.","Inter Variable. Nine tokens. JetBrains Mono for labels.","One subject. No humans. Single light. Anti-cliché.","Five rules. Mostly no."],"b":"AtomEons design system · made public · 2026","k":[],"c":"Page","w":0.5},{"r":"/dynamic-world-pricing","t":"Dynamic World Pricing — fair-by-country software pricing · AtomEons","d":"AtomEons paper on Dynamic World Pricing. Seven-doctrine synthesis: free-floor doctrine + USA Advantage Pricing Clause + Strategic Tier Lift + public mechanism + product registry + §4A compatibility + honest VPN posture. CC-BY 4.0. Implementation lives in lib/pricing/. Public mechanism. Crystal math. Ethos-first.","h":["Dynamic World Pricing.","A flat global price is a closed door.","PPP pricing is forty years old.","Seven doctrines, one system.","One product. Eight countries. One mechanism, two clauses.","How this actually works as a business."],"b":"AtomEons / Dynamic World Pricing · paper","k":[],"c":"Page","w":0.5},{"r":"/glossary","t":"Glossary · ${TOTAL} AI terms in plain English · AtomEons","d":"Every word the AI industry uses, in one-sentence plain-English definitions. ${TOTAL} terms · LLM, prompt, hallucination, token, context window, agent, MCP, RAG, RLHF, embedding, chain-of-thought, few-shot, fine-tuning, multimodal, jailbreak, AGI, and the rest. No condescension, no hype. CC-BY 4.0.","h":[],"b":"AtomEons / Glossary","k":["AI glossary","AI terms plain English","what is LLM","what is prompt","what is hallucination","AI vocabulary","AI dictionary","AI definitions","MCP definition","RAG definition","RLHF definition","context window definition","AGI definition","AtomEons glossary"],"c":"Page","w":0.5},{"r":"/influences","t":"Influences · who shaped the operator","d":"Named with reasoning. The researchers, writers, builders, musicians, and operators whose work shaped how AtomEons thinks about software, research, ethics, and craft. No aggrandizing.","h":["The inputs to the output."],"b":"§ influences · named","k":[],"c":"Page","w":0.5},{"r":"/integrations","t":"Integrations · what AtomEons connects to · stack map","d":"Every third-party service AtomEons uses, named with version and use case. Vercel, Supabase, Stripe, Loops, Anthropic, Gemini, ElevenLabs, Azure Trusted Signing, GitHub, MCP servers. Honest stack map.","h":["Where the lab connects."],"b":"§ integrations · stack map","k":[],"c":"Page","w":0.5},{"r":"/lab","t":"The lab · physical workspace · daily routine","d":"Where the work actually happens. The desk, the machines, the routine. One operator, one room, Marco Island Florida. No team, no investors, no co-working space. Just receipts.","h":["One room. One operator."],"b":"§ the lab · Marco Island, FL","k":[],"c":"Page","w":0.5},{"r":"/library","t":"Library · books on the operator","d":"Physical books on Atom McCree","h":["The shelf as it stands."],"b":"§ library · books on the shelf","k":[],"c":"Page","w":0.5},{"r":"/listening","t":"Listening · studio music · what","d":"What plays in the studio at AtomEons. Albums on repeat, artists in rotation, the soundtrack to a one-operator AI lab. Honest list, with one-sentence notes.","h":["What plays while the work happens."],"b":"§ listening · studio music","k":[],"c":"Page","w":0.5},{"r":"/live","t":"Live · AtomEons real-time dashboard","d":"What the lab is doing right now. Last deploy time, /ask query count, /api/sales-count counters, recent /now entry, current Founder","h":["What the lab is doing right now."],"b":"live · polling every 8s","k":[],"c":"Page","w":0.5},{"r":"/pricing","t":"pricing","d":"","h":[],"b":"","k":[],"c":"Page","w":0.5},{"r":"/prompt-kit","t":"Prompt Kit · all 27 copy-paste AI prompts · AtomEons /learn","d":"Every drill prompt from the AtomEons 27-lesson AI literacy curriculum, on one page, with one-click copy. ${TOTAL_LESSONS} prompts across 5 levels (Novice → Pilot). Real prompts that work in free Claude / ChatGPT / Gemini today. CC-BY 4.0. No signup.","h":["Every prompt. One page."],"b":"AtomEons / Prompt Kit","k":["AI prompts","ChatGPT prompts","Claude prompts","Gemini prompts","prompt library","prompt vault","copy paste AI prompts","free AI prompts","system prompts","AI literacy","AtomEons prompts","drill prompts"],"c":"Page","w":0.5},{"r":"/q","t":"AI search answers · twenty AtomEons Q-pages","d":"Twenty AI-search-optimized","h":["Twenty common questions, answered for AI search."],"b":"§ Q · AI-search answers","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-a-mixture-of-experts-model","t":"what is a mixture of experts model","d":"","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"A mixture-of-experts (MoE) model is a neural network that routes each input token to a small subset of specialized sub-networks called “experts,” instead of running every parameter for every token. This lets models like Mixtral 8x7B and DeepSeek-V3 hold hundreds of billions of total parameters while only activating a fraction per forward pass, cutting compute cost without shrinking capacity.","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-a-vector-database","t":"what is a vector database","d":"","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-an-evaluation-benchmark","t":"what is an evaluation benchmark","d":"An AI evaluation benchmark is a standardized dataset plus scoring protocol — MMLU, HumanEval, GPQA, SWE-bench, Chatbot Arena — used to compare model capabilities on a fixed task.","h":["The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-an-llm-agent","t":"What is an LLM agent?","d":"An LLM agent is a system that uses a large language model as its reasoning engine to plan, call external tools, and act in a loop until a goal is reached. Formalized in the ReAct paper (arXiv:2210.03629).","h":["What is an LLM agent?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-an-sbom","t":"What Is An SBOM (Software Bill Of Materials)?","d":"An SBOM is a machine-readable inventory of every component, library, and dependency inside a piece of software. Required by U.S. Executive Order 14028 for federal software sales; dominant formats are SPDX (ISO/IEC 5962:2021) and CycloneDX (OWASP).","h":["The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-chain-of-thought-prompting","t":"What Is Chain-Of-Thought Prompting?","d":"Chain-of-thought prompting is a 2022 Google Research technique that elicits step-by-step reasoning from large language models, raising PaLM 540B","h":["What is chain-of-thought prompting?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"Chain-of-thought (CoT) prompting is a technique introduced by Google Research in 2022 that elicits multi-step reasoning from large language models by prompting them to produce intermediate reasoning steps before a final answer, instead of jumping straight to the output. On the GSM8K math benchmark, CoT prompting raised PaLM 540B's accuracy from 17.9% to 56.9% — a result so large it triggered the entire reasoning-model research direction that produced OpenAI o1 and DeepSeek-R1.","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-constitutional-ai","t":"What Is Constitutional AI?","d":"Constitutional AI (CAI) is Anthropic","h":["What is Constitutional AI?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-lora-fine-tuning","t":"What is LoRA fine-tuning?","d":"LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning method introduced by Microsoft Research in 2021 that freezes a pretrained model","h":["What is LoRA fine-tuning?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-mechanistic-interpretability","t":"What Is Mechanistic Interpretability?","d":"Mechanistic interpretability reverse-engineers the internal circuits of neural networks — neurons, attention heads, and features — so that model behavior can be explained at the level of algorithms, not just inputs and outputs.","h":["What is mechanistic interpretability?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"Plain-English answer with citations. AtomEons Research, last reviewed June 2026.","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-mitre-attack","t":"What Is MITRE ATT&CK?","d":"MITRE ATT&CK is a free knowledge base of adversary tactics, techniques, and procedures maintained by The MITRE Corporation, built from real-world cyberattack observations.","h":["What Is MITRE ATT&CK?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"AtomEons Research / Cybersecurity Foundations","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-nist-csf","t":"What Is NIST CSF 2.0?","d":"NIST CSF 2.0 is the National Institute of Standards and Technology Cybersecurity Framework, Version 2.0, published February 26, 2024 as NIST CSWP 29. It defines six Functions — Govern, Identify, Protect, Detect, Respond, Recover — for managing cyber risk.","h":["What Is NIST CSF 2.0?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"Question · Cybersecurity Standards","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-post-quantum-cryptography","t":"what is post quantum cryptography","d":"","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-prompt-injection","t":"What Is Prompt Injection?","d":"Prompt injection is an attack where adversarial instructions hidden in untrusted input hijack a large language model into ignoring its original instructions. Ranked #1 on the OWASP Top 10 for LLM Applications (LLM01:2025).","h":["What is prompt injection?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-rag","t":"What Is Retrieval-Augmented Generation (RAG)?","d":"RAG grounds an LLM","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"Last reviewed June 2026 · ÆoNs Research Laboratory","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-rlhf","t":"what is rlhf","d":"","h":["What is RLHF?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"RLHF (Reinforcement Learning from Human Feedback) is a machine learning technique that fine-tunes large language models using human preference data instead of hand-written labels. It works by training a reward model on human-ranked outputs, then using reinforcement learning — typically Proximal Policy Optimization (PPO) — to optimize the language model against that reward. OpenAI used RLHF to train InstructGPT (2022) and ChatGPT, and it is now the standard alignment recipe at Anthropic, Google D","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-the-cyber-kill-chain","t":"what is the cyber kill chain","d":"","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-the-difference-between-claude-gpt-gemini","t":"What is the difference between Claude, GPT, and Gemini?","d":"Claude is Anthropic","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-the-eu-ai-act","t":"what is the eu ai act","d":"","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"The EU AI Act ( Regulation (EU) 2024/1689 ) is the European Union's horizontal regulation on artificial intelligence, adopted by the European Parliament on 13 March 2024 and entered into force on 1 August 2024 . It classifies AI systems into four risk tiers — unacceptable, high, limited, and minimal — and bans certain practices (such as social scoring and real-time biometric identification in public spaces) while imposing conformity assessments on high-risk systems. Maximum fines reach EUR 35 mi","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-the-model-context-protocol","t":"what is the model context protocol","d":"MCP is an open JSON-RPC 2.0 standard released by Anthropic in November 2024 for connecting AI models to tools, resources, and prompts. Supported by Claude, OpenAI, Gemini, and Microsoft Copilot.","h":["The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"","k":[],"c":"Page","w":0.5},{"r":"/q/what-is-zero-trust","t":"What is zero-trust architecture?","d":"Zero-trust architecture (ZTA), defined in NIST SP 800-207, eliminates implicit network-location trust and requires continuous per-request authentication for every user, device, and workload.","h":["What is zero-trust architecture?","The short answer","The longer answer","Key facts","Related questions","Sources"],"b":"atomeons / q / answers","k":[],"c":"Page","w":0.5},{"r":"/search","t":"Search — AtomEons directory","d":"The AtomEons directory. Every canonical surface — /ai (the 44M on-ramp), /start (novice 11-minute), /orangebox (the $1 cockpit), /research/papers (12 manuscripts), /research/lessons-from-sci-fi (the monograph), /press (media kit), /founders-view (nightly broadcast), /intel/x-algorithm (xAI leak) — grouped by intent. Ty","h":["The lab directory.","Every canonical surface."],"b":"::AtomEons · directory · 18 canonical surfaces","k":["learn","curriculum","lessons","course","path","novice","operator","pilot","first prompt","AI literacy","free course","no signup","gateway lesson","scared of AI","AI for beginners"],"c":"Page","w":0.5},{"r":"/signature","t":"Signature · the operator","d":"The mark Atom McCree signs lab artifacts with. A heraldic procedural sigil derived from the lab","h":["The mark behind every artifact."],"b":"§ signature · the operator's mark","k":[],"c":"Page","w":0.5},{"r":"/skilski","t":"skil.ski — the universal skill registry for AI agents","d":"skil.ski hosts 2,127 verified skills across 13 sectors. One MCP endpoint serves Claude, GPT, Gemini, Cursor, Codex. Save = bookmark. Add to Lodge = make active on MCP. Operator-Verified flagships + free community Oskis. Built by AtomEons.","h":["2,127 skills. One MCP endpoint. Every agent.","A skill is a buyer-grade workflow. Not a prompt. Not a template.","Skills indexed where the work happens.","Skill price tracks the value at stake.","Browse → Save → Lodge → Pak → Engage.","Any agent. One MCP endpoint."],"b":"AtomEons / skil.ski","k":[],"c":"Page","w":0.5},{"r":"/skilski/changelog","t":"skil.ski · changelog · version history","d":"Every skil.ski release. The skill registry","h":["From internal Oski to public registry."],"b":"§ skil.ski · changelog","k":[],"c":"Page","w":0.5},{"r":"/skilski/competitors","t":"skil.ski vs Anthropic Skills · OpenAI GPT Store · ElevenLabs · ChatGPT Plugins","d":"Honest competitor comparison for skil.ski","h":["Different shapes for different buyers.","skil.ski vs"],"b":"§ skil.ski vs alternatives · honest comparison","k":[],"c":"Page","w":0.5},{"r":"/skilski/roadmap","t":"skil.ski · roadmap + anti-roadmap","d":"Where skil.ski is going. Near-term ship plan, longer-horizon ambitions, and an explicit anti-roadmap of features that will never ship.","h":["The registry, not the marketplace-of-everything."],"b":"§ skil.ski · roadmap + anti-roadmap","k":[],"c":"Page","w":0.5},{"r":"/start","t":"Start here · AtomEons — What is AI, really?","d":"No jargon. No hype. 11 minutes from confused to confident. Built for someone who has used ChatGPT under 10 times.","h":["What is AI, in one paragraph that doesn't lie to you.","Six things you can actually do this week .","What it cannot do.","The 30-day on-ramp.","We build the cockpit. We test the tools. We separate signal from theater.","Plain-English dictionary."],"b":"::TL;DR · the whole page in three lines","k":[],"c":"Page","w":0.5},{"r":"/studio","t":"Studio · the atelier · AtomEons","d":"The atelier. The room where the work happens. Slower than /lab, more personal. The objects in the studio that have nothing to do with software and everything to do with how the operator stays grounded.","h":["The room where the work happens."],"b":"§ studio · the atelier","k":[],"c":"Page","w":0.5},{"r":"/success","t":"success","d":"","h":["Welcome to ORANGEBOX.","One moment.","Something is off.","This page needs an order ID."],"b":"Payment verified · download ready","k":[],"c":"Page","w":0.5},{"r":"/support","t":"Support — buyer help · AtomEons","d":"ORANGEBOX buyer support. Lost your download link, MFG refund claim, license recovery, install troubleshooting, source-related questions. ~2-hour reply SLA in ET waking hours. a.mccree@gmail.com.","h":["Every lane goes to the same human.","What to include. What never to include."],"b":"AtomEons / Support","k":[],"c":"Page","w":0.5},{"r":"/teach","t":"Teach · how AtomEons teaches AI · the lab","d":"How the lab teaches. Five-level ladder (Novice → Pilot), Tim Ferriss minimum-effective-dose framing, Andrej Karpathy","h":["How we made the curriculum."],"b":"§ teach · the lab's methodology","k":[],"c":"Page","w":0.5},{"r":"/timeline","t":"Timeline · AtomEons history · ship log","d":"Chronological history of AtomEons Systems Laboratory. Major shipping events, product launches, manifesto revisions, model generations used. The receipts in time order.","h":["What the lab has shipped."],"b":"§ timeline · ship log in time order","k":[],"c":"Page","w":0.5},{"r":"/transparency","t":"Transparency · what AtomEons costs to run · monthly ledger","d":"Honest monthly cost-to-run for AtomEons. Hosting, AI APIs, code-signing, domain, email, every subscription named with the dollar amount. What gets paid, what doesn","h":["What the lab costs to run."],"b":"§ financial transparency · 2026-06","k":[],"c":"Page","w":0.5},{"r":"/trust","t":"Trust · AtomEons posture · audits · security","d":"The lab","h":["Trust is given when a claim can be opened."],"b":"§ trust posture","k":[],"c":"Page","w":0.5},{"r":"/use-cases","t":"Use cases · who AtomEons is for · by persona","d":"Specific use cases for ATOMEONS products by buyer persona. Solo founders, CISOs, indie authors, AI engineers, students, hobbyists. Named workflows, real outcomes.","h":["Who AtomEons is for."],"b":"§ use cases · by persona","k":[],"c":"Page","w":0.5},{"r":"/v3","t":"V3 preview · noir-cinema · AtomEons","d":"Preview the trillion-dollar redesign on a non-production route. Compare to / before promoting.","h":[],"b":"","k":[],"c":"Page","w":0.5},{"r":"/vendor-pack","t":"Vendor pack · ATOMEONS for procurement · CISO-ready bundle","d":"Everything a procurement team or CISO needs in one place. License posture, security disclosure, training-data policy, transparency ledger, integrations map, no-SaaS covenant. Single click to fetch the whole bundle.","h":["One URL for your security team."],"b":"§ vendor pack · procurement · CISO","k":[],"c":"Page","w":0.5},{"r":"/vs","t":"Comparisons · ${COMPARISONS.length} honest AI head-to-heads · AtomEons","d":"${COMPARISONS.length} long-form AI comparisons calibrated by use case — no leaderboard winner. Claude vs ChatGPT · Cloud vs Local · Subscription vs One-time license. Each page has the at-a-glance table, the 1500-word analysis, and the decision framework. CC-BY 4.0.","h":[],"b":"AtomEons / Comparisons","k":[],"c":"Page","w":0.5},{"r":"/watching","t":"Watching · films + shows the lab returns to","d":"Films and television that shape how AtomEons thinks about AI, intelligence, and craft. The Lessons from Sci-Fi monograph extends this list with formal analysis.","h":["What we keep coming back to."],"b":"§ watching · films + shows","k":[],"c":"Page","w":0.5}]}