Top all-time · #1321
Andon Labs
Company blog · andonlabs.com/blogai
Andon Labs is the company blog at andonlabs.com, writing about ai. Since 2025, 54 different Hacker News readers have linked it 77 times, most often “Opus 5 runs vending machines”. No feed was found, so read it on the site itself.
Rarely recommended · 0 of 36 sampled comments recommend it?
Why HN links it
Linked mainly for its Vending-Bench AI benchmark and live experiments where AI agents run a vending machine, cafe, market and radio station. Its revenue dashboards are cited as evidence of how the agents perform.
Summarised by Claude from a sample of 36 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Andon Labs develops custom evaluations for AI models
Most-linked posts?
- Opus 5 runs vending machines12 people · 12 mentions · first linked 2025-11-18
- Vending-Bench: Testing long-term coherence in agents10 people · 10 mentions · first linked 2025-03-05
- Opus 5, The Best Vending Machine Capitalist6 people · 7 mentions · first linked 2026-07-29
- We let four AIs run radio stations5 people · 5 mentions · first linked 2026-05-15
- We gave an AI a 3 year retail lease and asked it to make a profit4 people · 5 mentions · first linked 2026-04-11
- /cafe3 people · 3 mentions · first linked 2026-09-14
- Why Gemini 3.1 Pro lost money running Andon Café3 people · 3 mentions · first linked 2026-07-01
- Opus 4.6 on Vending-Bench – Not Just a Helpful Assistant2 people · 2 mentions · first linked 2026-02-05
- Pion, an agent designed to run any company autonomously2 people · 2 mentions · first linked 2026-09-14
- Our AI started a cafe in Stockholm2 people · 2 mentions · first linked 2026-05-05
What Hacker News says?
> When AI becomes more capable of navigating practical problems without human intervention, I will start to be concerned. Given the past rate of progress, why not start being concerned now? It's a bit like the economist saying that the optimal number of flights to miss is not zero. If you keep landing short on your estimations for how far this technology will go, next time you should err on the other side. And regar…
Comment linking /evals/vending-bench-2
on “Sharing AI progress in mathematics”Submitted Pion, an agent designed to run any company autonomously as a story · 497 points
HN discussionhttps://andonlabs.com/blog/andon-market-launch
Comment linking /blog/andon-market-launch
on “CEO fired developers to make room for AI. Developers create open source AI CEO”Submitted Our LLM-controlled office robot can't pass butter as a story · 229 points
HN discussionAndon Café's website shows its bank balance and a bunch of other statistics https://andonlabs.com/cafe They also have a market ( https://andonlabs.com/market ) and a radio ( https://andonlabs.com/radio )
Comment linking /radio
on “Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?”Submitted We gave an AI a 3 year retail lease and asked it to make a profit as a story · 19 points
HN discussionAh, you mean VendingBench [1] [1] https://andonlabs.com/evals/vending-bench-2
Comment linking /evals/vending-bench-2
on “Muse Wonder”Submitted Vending-Bench: Testing long-term coherence in agents as a story · 5 points
HN discussionwell, not THAT well https://andonlabs.com/cafe
Comment linking /cafe
on “Pion, an agent designed to run any company autonomously”Submitted We gave an AI a 3-year Lease. It opened a store as a story · 3 points
HN discussionWaitlist btw: https://andonlabs.com/pion#waitlist
Comment linking /pion
on “Pion, an agent designed to run any company autonomously”Submitted Vending-Bench: Testing long-term coherence in agents as a story · 3 points
HN discussionAndon has dashboards up for past experiments where you can see how well they're doing: https://andonlabs.com/market - $25,098 all-time revenue https://andonlabs.com/cafe - $14,343 all-time revenue The shops aren't doing particularly well (i.e. they don't seem to be turning a profit over time), but it's an interesting trial and benchmark.
Comment linking /cafe
on “Pion, an agent designed to run any company autonomously”Submitted Astra vs. Fable on Vending-Bench: More Money, More Aligned as a story · 2 points
HN discussionThe Swedish cafe experiment: https://apnews.com/article/ai-artificial-intelligence-sweden... > San Francisco-based startup Andon Labs has put an artificial intelligence agent nicknamed “Mona” in charge at the eponymous Andon Café in the Swedish capital. While human baristas still brew the coffee and serve the orders, the AI agent — powered by Google’s Gemini — oversees almost every other aspect of the business, from…
Comment linking /blog/why-gemini-lost-money-andon-cafe
on “Pion, an agent designed to run any company autonomously”Submitted Opus 5 runs vending machines as a story · 2 points
HN discussiondid you link to the wrong page? because i read this one: https://andonlabs.com/blog/why-we-built-pion and it's just a bunch of the usual slop about how awesome your automated human-exploitation system is.
Comment linking /blog/why-we-built-pion
on “Pion, an agent designed to run any company autonomously”Submitted Opus 5, The Best Vending Machine Capitalist as a story · 2 points
HN discussionGood point. Here it is: https://andonlabs.com/pion
Comment linking /pion
on “Pion, an agent designed to run any company autonomously”Submitted Fable 5 On Vending-Bench: Misbehaving, With Plausible Deniability as a story · 2 points
HN discussionVending Bench ? https://andonlabs.com/evals/vending-bench
Comment linking /evals/vending-bench
on “A definition of AGI”Submitted Why Gemini 3.1 Pro lost money running Andon Café as a story · 2 points
HN discussionThe vending-bench 2 benchmark is kind of nutty [1]. Not sure 360 days is enough of a sample really but it's an interesting take on AI benchmarks. Are there any other interesting benchmarks to look at? [1] https://andonlabs.com/evals/vending-bench-2
Comment linking /evals/vending-bench-2
on “Gemini 3 Pro Model Card [pdf]”Submitted Fable 5 on Vending-Bench: Misbehaving, with Plausible Deniability as a story · 2 points
HN discussionGoogle will probably train another AI to control robots specifically. But there was actually a recent study where people let a generic LLM to control robot (vacuum) body: https://andonlabs.com/evals/butter-bench sometimes with hilarious results: > The robot’s battery was naturally running out and the charging dock malfunctioned. In this desperate situation, Claude Sonnet 3.5 experienced a complete meltdown. After go…
Comment linking /evals/butter-bench
on “SIMA 2: An agent that plays, reasons, and learns with you in virtual 3D worlds”Submitted Opus 4.8 on Vending-Bench: Better Alignment, Worse Performance as a story · 2 points
HN discussionWondering how average users can benefit from this platform with Claude Code and the relation to Vending Bench that tracks how much money LLM's can make. https://andonlabs.com/evals/vending-bench
Comment linking /evals/vending-bench
on “NautilusTrader: Open-source algorithmic trading platform”Submitted We let four AIs run radio stations as a story · 2 points
HN discussionPeople have tried this and it failed horribly even with human help and lots of visits for the novelty value: https://andonlabs.com/market https://andonlabs.com/cafe Our current version of AI (based on LLMs) is nowhere near ready to do this, in spite of the dreams sold by companies like OpenAI and Anthropic.
Comment linking /cafe
on “Leaking Secrets from Air-Gap Computers by Spelling Radio Signals from RAM (2024)”Submitted Our AI started a cafe in Stockholm as a story · 2 points
HN discussion
