Sources
The course was distilled from 62 pages of vendor docs and cookbooks, 58 talks and podcasts, 18 papers, and 61 articles, repositories and discussions. Quotes in the lessons link to the exact moment in each talk. Not affiliated with or endorsed by anyone listed.
ArticleHow to build typed decisions with JevFlavio CopesCited in lesson 1, lesson 3, lesson 4, lesson 5, lesson 7, lesson 10, lesson 13, lesson 19ArticleIntroducing System One Models & JevDiogo AlmeidaCited in lesson 1, lesson 2, lesson 3, lesson 12, lesson 16, lesson 17, lesson 18DocsJev 1.13 jaggednessTypeSafe AI docsCited in lesson 6, lesson 7, lesson 8, lesson 15, lesson 16, lesson 19, lesson 22CodeTypeSafe agent skill (SKILL.md)TypeSafe AICited in lesson 8, lesson 12, lesson 13, lesson 15, lesson 18, lesson 22TalkAsking JEV's Creator the Hard Questions About This Hot New AI ModelThe TWIML AI Podcast with Sam CharringtonCited in lesson 3, lesson 9, lesson 10, lesson 17, lesson 19PaperCalibrated Decisions at Scale: Converting Police Crash Narratives into Probabilistic Crash Variables with…Amir Rafe amir rafe eduCited in lesson 6, lesson 9, lesson 11, lesson 18, lesson 20DiscussionHacker News: Introducing System One Models and JevHacker News commentersCited in lesson 3, lesson 5, lesson 11, lesson 18, lesson 19ArticleJev introduces a new shape of LLM—System One, aka Decision ModelsSimon WillisonCited in lesson 1, lesson 5, lesson 12, lesson 15, lesson 21ArticleThe Model That Doesn't TalkDaniel SamCited in lesson 1, lesson 2, lesson 3, lesson 6, lesson 10Talk10 Levels of Jev For Agentic EngineersIndyDevDanCited in lesson 5, lesson 16, lesson 17, lesson 20ArticleAn early-access test of TypeSafe's Jev: calibrated judgments for half a centEmil LindforsCited in lesson 5, lesson 11, lesson 18, lesson 20DocsConfidenceTypeSafe AI docsCited in lesson 6, lesson 7, lesson 10, lesson 11DocsHow to build with TypeSafeTypeSafe AI docsCited in lesson 1, lesson 6, lesson 8, lesson 12TalkLLM Evals: Common MistakesHamel HusainCited in lesson 5, lesson 7, lesson 20, lesson 21Codestrands-decider (README)AWS Strands LabsCited in lesson 3, lesson 4, lesson 6, lesson 17ArticleThe OpenAI Decisions API Needs a Confidence You Can Trust | AnthusRyan PorterCited in lesson 9, lesson 10, lesson 19, lesson 20ArticleUsing LLM-as-a-Judge For Evaluation: A Complete Guide – Hamel’s BlogHamel HusainCited in lesson 5, lesson 7, lesson 16, lesson 20DocsDecisions (OpenAI API guide)OpenAI docsCited in lesson 1, lesson 4, lesson 7ArticleDocument Review With Unsiloed and TypeSafe JevUnsiloed AICited in lesson 5, lesson 16, lesson 21ArticleJev and System One models - a semantic if statement for your codeSebastien DuboisCited in lesson 14, lesson 19, lesson 21ArticleJev is the 🐠 at the poker tableMike RamosCited in lesson 12, lesson 18, lesson 19TalkOlivier Sibony: How can you overcome noise and bias in decision making?CFA InstituteCited in lesson 7, lesson 12, lesson 21DocsPrimitives (Questions)TypeSafe AI docsCited in lesson 5, lesson 8, lesson 13TalkThinking, Fast and Slow | Daniel Kahneman | Talks at GoogleTalks at GoogleCited in lesson 2, lesson 9, lesson 12ArticleApgar score - WikipediaAuthority control databases International GND FAST National United States Israel Other Yale LUXCited in lesson 12, lesson 21TalkBuilding Decision Agents with LLMs & Machine Learning ModelsIBM TechnologyCited in lesson 10, lesson 21ArticleBuilding Effective AI AgentsAnthropicAICited in lesson 14, lesson 17TalkCloudflare Just Challenged Jev, So I Tested Their New Model (Clef)Better StackCited in lesson 19, lesson 20Docscloudflare/clef model cardCloudflareCited in lesson 3, lesson 4DocsComposite scoringTypeSafe AI docsCited in lesson 12, lesson 18TalkDaniel Kahneman: Thinking Fast and Slow, Deep Learning, and AI | Lex Fridman Podcast #65Lex FridmanCited in lesson 2, lesson 21TalkDr. Desi Ivanova | Estimating and Calibrating Uncertainty in LLMsINI Seminar Room 1Cited in lesson 3, lesson 9Codefirelex/jeff: Millisecond decisions, any domain: a 0.8B open "System 1" model that picks between your…firelex/jeffCited in lesson 14, lesson 19Codejaredpalmer/kev: Jev-like family of decision models built on top of Qwen3.5/3.8 you can train and run on…jaredpalmer/kevCited in lesson 4, lesson 11DocsGuardrails for LLMsTypeSafe AI docsCited in lesson 16, lesson 22DocsHierarchical classificationTypeSafe AI docsCited in lesson 13, lesson 15TalkHow to Cut Your LLM Classification Costs by 90%Hamel HusainCited in lesson 11, lesson 14DocsIntent routingTypeSafe AI docsCited in lesson 14, lesson 22DocsIntroducing Clef: Cloudflare's first open-source decision models, now on Workers AICloudflare docsCited in lesson 1, lesson 4ArticleIntroducing Clef: our open-source decision models, and new RL fine-tuning platformCloudflareCited in lesson 3, lesson 4ArticleIntroducing Strands Decider 2B: a small, open source, decision modelMarc Brooker; Mike Chambers; Fabio Nonato de PaulaCited in lesson 1, lesson 17DocsIntroductionTypeSafe AI docsCited in lesson 1, lesson 2TalkJev Explained: Demos and Use CasesSyntaxCited in lesson 1, lesson 14Talkjev: on | Diogo Almeida founder of Typesafe AI, Runtime 2026ModalCited in lesson 2, lesson 17PodcastJev: System One models for Prod, not God (Latent Space podcast)Diogo Almeida with swyxCited in lesson 2, lesson 8CodeJevBench v1 (benchmark for Jev-class typed decision models)JevBench authorsCited in lesson 4, lesson 20ArticleLies, Damned Lies, and BenchmarksTypeSafe AICited in lesson 4, lesson 20DocsModelsTypeSafe AI docsCited in lesson 4, lesson 8DocsNoulTypeSafe AI docsCited in lesson 5, lesson 8DocsParallel questionsTypeSafe AI docsCited in lesson 13, lesson 21PaperPinocchio: Fast Uncertainty Estimates for Black-Box Language ModelsKevin David HayesCited in lesson 9, lesson 11DocsPre-parsed value extractionTypeSafe AI docsCited in lesson 15, lesson 22DocsSkill suggestionTypeSafe AI docsCited in lesson 13, lesson 17DocsSpeculative fan-outTypeSafe AI docsCited in lesson 13, lesson 22ArticleStrands Decider: Why Not an Encoder?Marc BrookerCited in lesson 3, lesson 13DocsStructure recoveryTypeSafe AI docsCited in lesson 13, lesson 15ArticleTypeSafe AI Releases Jev: a Decision-Only Model That Returns Typed ProbabilitiesInfoQCited in lesson 10, lesson 19TalkWhat Is Jev? The AI Model That Doesn't Generate TextIBM TechnologyCited in lesson 9, lesson 10Talk[1hr Talk] Intro to Large Language ModelsAndrej KarpathyCited in lesson 2Talk$1 AI Guardrails: The Unreasonable Effectiveness of Finetuned ModernBERTs – Diego CarpinteroAI EngineerCited in lesson 16DocsAI primerTypeSafe AI docsCited in lesson 9ArticleAn Independent Evaluation of TypeSafe's Jev | Vals AIConnor FrankCited in lesson 20DocsAutoresearch feature discoveryTypeSafe AI docsCited in lesson 18ArticleAWS launches a local answer to TypeSafe's Jev decision modelThe New StackCited in lesson 4ArticleBrier score - WikipediaWikimedia Foundation, Inc.Cited in lesson 9DocsChoiceTypeSafe AI docsCited in lesson 6DocsClassification using confidenceTypeSafe AI docsCited in lesson 11DocsConfidence-gated routingTypeSafe AI docsCited in lesson 10DocsDate extractionTypeSafe AI docsCited in lesson 15TalkDecision Hygiene for Reducing Bias and Noise in Judgment | Daniel Kahneman & Robinson ErhardtRobinson's Podcast ClipsCited in lesson 21TalkDiogo Almeida - TypeSafe AIFounders You Should KnowCited in lesson 1DocsDouble-checking citationsTypeSafe AI docsCited in lesson 16ArticleDual process theory - WikipediaWikimedia Foundation, Inc.Cited in lesson 2Paperelkan-cost-sensitiveCited in lesson 10TalkFake Jev demos are taking over the internetSteve (Builder.io)Cited in lesson 17PaperFrugalGPT: How to Use Large Language Models While Reducing Cost and Improving PerformanceChen, Zaharia, ZouCited in lesson 14PaperGPT-4 Technical ReportOpenAI et al.Cited in lesson 9DiscussionHacker News: Decisions API is in public betaHacker News commentersCited in lesson 14DiscussionHacker News: Jev in 25 Lines of PythonHacker News commentersCited in lesson 14DiscussionHacker News: Strands Decider 2B: a small, open-source, decision modelHacker News commentersCited in lesson 17TalkHow Jev Turns AI Into Software That Gets Things Donea16zCited in lesson 1TalkHow We Build Effective Agents: Barry Zhang, AnthropicAI EngineerCited in lesson 14TalkJev - The Ultimate Classification Model?Sam WitteveenCited in lesson 12TalkJev CEO: I made ChatGPT, now I'm building what's nextAI EngineerCited in lesson 3ArticleJev in 25 lines of Python - NobodyWhoDuarte O CarmoCited in lesson 3TalkJev is incredibleTheo - t3․ggCited in lesson 19ArticleJev-class models — JevBench v1.0Benchmark HeavenCited in lesson 4ArticleJevMan: a Pac-Man benchmark for decision models | Opper AIOpper AICited in lesson 17ArticleJevons paradox - WikipediaWikimedia Foundation, Inc.Cited in lesson 18PaperJudging LLM-as-a-Judge with MT-Bench and Chatbot ArenaZheng et al.Cited in lesson 6Paperkahneman-nobelCited in lesson 2PaperLanguage Models (Mostly) Know What They KnowKadavath et al.Cited in lesson 9PaperLarge Language Models Sensitivity to The Order of Options in Multiple-Choice QuestionsPezeshkpour, HruschkaCited in lesson 6ArticleLaya — 33ms Multilingual System 1 Decision EngineCited in lesson 10DocsLine-by-line searchTypeSafe AI docsCited in lesson 15PaperOn Calibration of Modern Neural NetworksGuo et al.Cited in lesson 9ArticleOpenAI answers TypeSafe's Jev with a Decision API built on LunaFrederic LardinoisCited in lesson 4TalkOpenAI's Decisions API just dropped. Here's how it compares to Jev.Mark KashefCited in lesson 4TalkProbability Calibration : Data Science ConceptsritvikmathCited in lesson 9DocsRe-rankingTypeSafe AI docsCited in lesson 15PaperRouteLLM: Learning to Route LLMs with Preference DataOng et al.Cited in lesson 14DocsScoreTypeSafe AI docsCited in lesson 7DocsSDE cascadeTypeSafe AI docsCited in lesson 14PaperSelective Classification for Deep Neural NetworksGeifman, El-YanivCited in lesson 11DocsStateTypeSafe AI docsCited in lesson 8TalkStatistical Prediction Rules (SPRs) vs. Clinical expertsDailyDataCited in lesson 12Codestrands-decider: agent exampleAWS Strands LabsCited in lesson 17Codestrands-decider: evaluationAWS Strands LabsCited in lesson 5Codestrands-decider: inferenceAWS Strands LabsCited in lesson 13TalkSupercharging LLM Classifications with LogprobsArcturus LabsCited in lesson 3ArticleThe lethal trifecta for AI agents: private data, untrusted content, and external communicationSimon WillisonCited in lesson 16TalkUsing Jev In Your Agent HarnessSam WitteveenCited in lesson 2TalkWe tested JevBench's top three against Jev on 3,080 bank messagesModel FatigueCited in lesson 11TalkWhy AI evals are the hottest new skill for product builders | Hamel Husain & Shreya ShankarLenny's PodcastCited in lesson 20TalkWhy We Made Jev — Diogo Almeida, TypeSafe Co-founder & CEOKi-jung Kim (akaslany)Cited in lesson 8TalkYoshua Bengio | From System 1 Deep Learning to System 2 Deep Learning | NeurIPS 2019Preserve KnowledgeCited in lesson 2PaperA Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty QuantificationAngelopoulos, BatesTalkA Tutorial on Conformal PredictionAnastasios Nikolas AngelopoulosDocsAdvanced usageTypeSafe AI docsDocsAdvanced: structureTypeSafe AI docsDocsAgent skillTypeSafe AI docsDocsAPI referenceTypeSafe AI docsDocsAPI referenceTypeSafe AI docsDocsAPI referenceTypeSafe AI docsDocsBasic usageTypeSafe AI docsDocsBoxTypeSafe AI docsArticleBuild your own decision modelNish TahirTalkBuilding a Harness with JevLangChainPaperCalibrate Before Use: Improving Few-Shot Performance of Language ModelsZhao et al.PaperCalibrated Decisions at Scale: police crash narratives with a System One model (abstract)arXivPaperCan LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMsXiong et al.DocsChangelogTypeSafe AI docsDocsChangelogTypeSafe AI docsDocsClassifying RAG passagesTypeSafe AI docsDocsClient SDKsTypeSafe AI docsArticleComposable AI: Build Prod, Not God (manifesto)TypeSafe AIDocsCookbooksTypeSafe AI docsTalkDaniel Kahneman and Yann LeCun: How To Get AI To Think Like Humans (Full Episode)Alex KantrowitzDocsDemosTypeSafe AI docsDocsExample use casesTypeSafe AI docsTalkFrugalGPT: Reduce Cost and Improve Performance of Large Language Models (James Zou, PhD)Synthetic Intelligence ForumDocsFunction callingTypeSafe AI docsCodePostHog/jeeves: Jeeves – Reasoning improves Jev-like decision modelsPostHog/jeevesDiscussionHacker News: Clef: Open-weight decision models, and new RL fine-tuning platformHacker News commentersTalkHow to Measure LLM Confidence: Logprobs & Structured OutputNeuralNineTalkHow to train INSANELY fast JEV Decision Models from scratch (architecture deep-dive)Neural Breakdown with AVBTalkI Collected 562 Jev Demos So You Don’t Have ToWeb Dev CodyTalkI Tested Jev (TypeSafe AI) in Production: An Honest ReviewNico ManzanequeTalkI Tested Jev vs 12 Local Decision Models. Here's What I’d Use...The AI AutomatorsDocsImagesTypeSafe AI docsTalkIntroducing the Decisions APIOpenAIDocsJavaScript SDKTypeSafe AI docsTalkJev explained in 7min..Caleb Writes CodeTalkJev Explained: The Fast AI Model We Tried to Break on PurposeBoundary and HumanLayerTalkJev Sucks, ActuallyDuncan Rogoff | Learn Claude CodeTalkJev vs OpenAI Decisions API: I Ran 505 Real Tests (Use This One)Hyperautomation LabsTalkJev vs. Clef: Which decision model works better in practice?Rafael Quintanilha – QuantBrasilDocsJev with coding agentsTypeSafe AI docsTalkJev: ChatGPT Co-Creator’s Answer to RLHFAI CouncilArticleJev: For the builders AND dreamersTypeSafe AICodeJevBench by Benchmark HeavenBenchmark HeavenPaperJust Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models…Tian et al.DocsKnowledge graph entity alignmentTypeSafe AI docsDocsLegalTypeSafe AI docsArticleOllaya · Run decision models locally.OllayaTalkOpenAI DevDay 2026 Keynote (FULL)OpenAIDocsPatternsTypeSafe AI docsArticlePatterns for Building LLM-based Systems & ProductsEugene YanTalkPydantic is all you need: Jason LiuAI EngineerDocsQuick startTypeSafe AI docsArticleRelease: llm-typesafe 0.1a0Simon WillisonDocsSelf-consistency: choicesTypeSafe AI docsDocsSelf-consistency: noulsTypeSafe AI docsDocsSimple memory systemTypeSafe AI docsDocsSmart home assistant demoTypeSafe AI docsTalkStop Asking LLMs for Confidence Scores (Do this instead)Coding Crash CoursesTalkStop Overpaying for Intelligence | DevDay 2026OpenAIDocsstrands-decider-2B model cardAWS Strands LabsCodestrands-decider: architectureAWS Strands LabsCodestrands-decider: JevBench notesAWS Strands LabsCodestrands-decider: resultsAWS Strands LabsCodestrands-decider: visionAWS Strands LabsDocsSystem OneTypeSafe AI docsArticleThe Bitter LessonArticleThe Bitterest LessonDiogo AlmeidaPaperThinking Fast and Slow in AIBooch et al.DocsTwitch chat filterTypeSafe AI docsArticleTypeSafe A raises Series AITypeSafe AITalkTypeSafe Jev Fact-Check: The 200x AI Hype vs What's ProvenPrism LabsArticleTypeSafe launched Jev because sequential LLMs are "totally useless for computers"The New StackDocsTypeSafe Python SDKTypeSafe AI docsArticleTypeSafe's Jev Can't See. I Made It Guess What I Drew Anyway – Bartosz Mikulski | AI & MLOps EngineerBartosz MikulskiArticleUsing logprobsOpenAI DevelopersTalkWe need to talk about Jev...Matthew BermanTalkWhat is Jev? (and is laya better?)GitButlerTalkWhen calibration beats metricsProbablArticleWhy Dwarkesh is Wrong about Computer Use + How OpenAI shipped its Jev competitor in 1 WeekLatent SpaceTalkWhy I switched my RAG decisions to Jev...The AI AutomatorsArticleWill OpenAI Eat Jev's Lunch?arcturus-labs.com