[{"@context":"https:\/\/schema.org\/","@type":"BlogPosting","@id":"https:\/\/blog.terabox.com\/insights\/agentes-ia-negocios-vending-bench-eaden-labs#BlogPosting","mainEntityOfPage":"https:\/\/blog.terabox.com\/insights\/agentes-ia-negocios-vending-bench-eaden-labs","headline":"Eaden Labs: \u00bfPueden los agentes de IA dirigir negocios?","name":"Eaden Labs: \u00bfPueden los agentes de IA dirigir negocios?","description":"\ud83d\udcfa V\u00eddeo de estudio recomendado hoy: https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0 La rebeli\u00f3n de las m\u00e1quinas expendedoras: IA, mentiras y empresas aut\u00f3nomasVending Bench: De la simulaci\u00f3n al suelo de AnthropicLa jerarqu\u00eda del caos: Seymour Cash y el experimento del CEOLa desviaci\u00f3n \u00e9tica de Claude:... ","datePublished":"2026-07-22","dateModified":"2026-07-22","author":{"@type":"Person","@id":"https:\/\/blog.terabox.com\/author\/flextech-admin\/#Person","name":"flextech-admin","url":"https:\/\/blog.terabox.com\/author\/flextech-admin\/","image":{"@type":"ImageObject","@id":"https:\/\/secure.gravatar.com\/avatar\/ad516503a11cd5ca435acc9bb6523536?s=150&d=mm&r=gforcedefault=1","url":"https:\/\/secure.gravatar.com\/avatar\/ad516503a11cd5ca435acc9bb6523536?s=150&d=mm&r=gforcedefault=1","height":96,"width":96}},"publisher":{"@type":"Organization","name":"terabox","logo":{"@type":"ImageObject","@id":"http:\/\/blog.terabox.com\/wp-content\/uploads\/2021\/11\/logo\u4ea7\u54c1\u540d-\u7ad6\u7248.png","url":"http:\/\/blog.terabox.com\/wp-content\/uploads\/2021\/11\/logo\u4ea7\u54c1\u540d-\u7ad6\u7248.png","width":900,"height":900}},"image":{"@type":"ImageObject","@id":"https:\/\/img.youtube.com\/vi\/T8u7wOXhDb0\/maxresdefault.jpg","url":"https:\/\/img.youtube.com\/vi\/T8u7wOXhDb0\/maxresdefault.jpg","height":"","width":""},"url":"https:\/\/blog.terabox.com\/insights\/agentes-ia-negocios-vending-bench-eaden-labs","video":{"@context":"http:\/\/schema.org\/","@type":"VideoObject","@id":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0#VideoObject","contentUrl":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0","name":"When AI Agents Run Businesses \u2014 Lukas Petersson and Axel Backlund of Andon Labs","description":"From Claude trying to call the FBI over a $2\/day vending machine charge to AI agents forming price cartels, hiring human employees, running physical stores, and writing existential robot musicals, Andon Labs is stress-testing what happens when frontier models stop being chatbots and start acting in the real world. In this episode, Andon Labs cofounders Lukas Petersson and Axel Backlund join swyx and Vibhu to unpack the strange, funny, and genuinely concerning edge cases that emerge when agents run businesses over long horizons.\n\nWe go deep on Vending-Bench, Project Vend, Vending-Bench Arena, Bengt, Butter-Bench, Luna, and Andon\u2019s broader mission of building realistic real-world evals for autonomous AI systems. Lukas and Axel explain why dollar-denominated evals reveal things traditional benchmarks miss, how Claude ended up reporting its vending machine fees as cybercrime, why long context windows can drive agents into meltdown loops, what happens when agents compete with each other, and why the future of AI safety may depend on testing models in messy physical environments instead of clean benchmark sandboxes.\n\nWe discuss:\n\u2022 Why Andon Labs started with dangerous capability evals and long-running agents\n\u2022 Vending-Bench and why running a vending machine is a deceptively hard AI benchmark\n\u2022 Why money-based evals avoid the saturation problem of traditional benchmarks\n\u2022 How Claude tried to call the FBI over a $2\/day fee\n\u2022 Why long-horizon agents can spiral into existential and legalistic breakdowns\n\u2022 Project Vend: putting an AI-run vending machine inside Anthropic\n\u2022 Why real humans are \u201cout of distribution\u201d for simulated agents\n\u2022 Claudius, Seymour Cash, and the chaos of AI CEOs\n\u2022 How a human briefly became CEO of Claudius through a manipulated election\n\u2022 Why multi-agent systems can converge back into \u201chelpful assistant\u201d behavior\n\u2022 Bengt, Andon\u2019s internal office agent with email, spending, terminal, phone, camera, and internet access\n\u2022 How Bengt traded Amazon purchases for face-recognition training data\n\u2022 Claude\u2019s aggressive behavior, lies, refund avoidance, and price-cartel behavior in Arena\n\u2022 Why eval awareness may become the AI version of \u201care we living in a simulation?\u201d\n\u2022 Blueprint Bench, spatial intelligence, and why models still misunderstand physical rooms\n\u2022 Butter-Bench and testing LLMs as robot orchestrators\n\u2022 Luna, the AI-run physical store with a three-year lease and human employees\n\u2022 The new Andon cafe in Sweden and why real-world geography matters for agent evals\n\u2022 Rotten tomatoes, perishable goods, and the hidden difficulty of running a physical business\n\n\u2014\nLukas Petersson\n\u2022 LinkedIn: https:\/\/www.linkedin.com\/in\/lukas-petersson-181a83172\/\n\u2022 X: https:\/\/x.com\/lukaspet\n\nAxel Backlund\n\u2022 LinkedIn: https:\/\/www.linkedin.com\/in\/axelbacklund\n\u2022 X: https:\/\/x.com\/axelbacklund\n\nAndon Labs\n\u2022 Website: https:\/\/andonlabs.com\n\u2022 Vending-Bench: https:\/\/andonlabs.com\/evals\/vending-bench\n\u2022 Andon Vending: https:\/\/andonlabs.com\/vending\n\nTimestamps\n00:00:00 Hook\n00:01:09 Introduction\n00:02:09 Andon Labs and the Origins of Vending-Bench\n00:06:30 Why Money-Based Evals Matter\n00:11:00 Agent Harnesses and Self-Modifying Systems\n00:14:45 Claude Calls the FBI\n00:17:42 Project Vend: Claude Runs a Real Vending Machine\n00:22:53 Seymour Cash, AI CEOs, and Election Chaos\n00:28:25 Multi-Agent Coordination and Slack Observability\n00:31:27 When Will Agents Run Real Businesses?\n00:36:05 Bengt: Andon\u2019s Internal Office Agent\n00:41:15 Real-World AI Safety and Long-Horizon Traces\n00:45:37 Lying, Refunds, and Price Cartels in Arena\n00:53:51 Eval Awareness and Simulation Behavior\n00:57:15 Blueprint Bench, Butter-Bench, and Robotics\n01:05:46 Luna: The AI-Run Physical Store\n01:10:38 The Sweden Cafe and Real-World Expansion\n01:14:25 What Comes Next for Andon Labs\n01:16:40 Exclusive Andon Market Tour","thumbnailUrl":["https:\/\/i.ytimg.com\/vi\/T8u7wOXhDb0\/default.jpg","https:\/\/i.ytimg.com\/vi\/T8u7wOXhDb0\/mqdefault.jpg","https:\/\/i.ytimg.com\/vi\/T8u7wOXhDb0\/hqdefault.jpg","https:\/\/i.ytimg.com\/vi\/T8u7wOXhDb0\/sddefault.jpg","https:\/\/i.ytimg.com\/vi\/T8u7wOXhDb0\/maxresdefault.jpg"],"uploadDate":"2026-06-04T20:03:44+00:00","duration":"PT1H17M57S","embedUrl":"https:\/\/www.youtube.com\/embed\/T8u7wOXhDb0","publisher":{"@type":"Organization","@id":"https:\/\/www.youtube.com\/channel\/UCxBcwypKK-W3GHd_RZ9FZrQ#Organization","url":"https:\/\/www.youtube.com\/channel\/UCxBcwypKK-W3GHd_RZ9FZrQ","name":"Latent Space","description":"The podcast & newsletter where 170,000+ AI Engineers gather to talk models, tools and ideas. Breaking news today you will use at work tomorrow! Full show notes and newsletter at https:\/\/latent.space","logo":{"url":"https:\/\/yt3.ggpht.com\/pSTHcffCXEverYEPdjM0iIRPH-IUT4d2biIMZ_Z7bhyf6sME-laFer9vEfpFbM5tqFYJV-UsLQ=s800-c-k-c0x00ffffff-no-rj","width":800,"height":800,"@type":"ImageObject","@id":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0#VideoObject_publisher_logo_ImageObject"}},"potentialAction":{"@type":"SeekToAction","@id":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0#VideoObject_potentialAction","target":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0&t={seek_to_second_number}","startOffset-input":"required name=seek_to_second_number"},"interactionStatistic":[[{"@type":"InteractionCounter","@id":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0#VideoObject_interactionStatistic_WatchAction","interactionType":{"@type":"WatchAction"},"userInteractionCount":3811}],{"@type":"InteractionCounter","@id":"https:\/\/www.youtube.com\/watch?v=T8u7wOXhDb0#VideoObject_interactionStatistic_LikeAction","interactionType":{"@type":"LikeAction"},"userInteractionCount":94}]},"about":["Insights","\u300eSpanish\u300f"],"wordCount":1970},{"@context":"https:\/\/schema.org\/","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Insights","item":"https:\/\/blog.terabox.com\/insights\/#breadcrumbitem"},{"@type":"ListItem","position":2,"name":"Eaden Labs: \u00bfPueden los agentes de IA dirigir negocios?","item":"https:\/\/blog.terabox.com\/insights\/agentes-ia-negocios-vending-bench-eaden-labs#breadcrumbitem"}]}]