Lukas Petersson
@lukaspet
Vending machines @andonlabs
Behind the scenes of Project Vend! In this special episode of Audio Tokens, we go deeper into Project Vend, the autonomous vending machine @andonlabs put in @AnthropicAI 's office. Daniel Freeman and @axelbacklund share unreleased anecdotes and ask questions like: Is this good…
I do not see any tungsten cubes in there, so far Claudius is winning
Can't wait
Just left the @Tesla design studio. Most epic demo ever by end of year. Ever.
Turns out @elonmusk likes vending machines, and his AI is good at managing them
Grok 4 on Vending Bench Grok 4 gets the #1 spot. Double the net worth of Claude Opus 4.
Let's put Claudius on it.
i used to eat sushi here every day when i lived in sweden and then they went bankrupt when i moved :(
Project Press next?
Media seems to be excited about Project Vend! But wouldn't count Claude out as a manager despite the headlines. Just give us a few more iterations 🛤️
Buy our vending machine and turn your daily life into an episode of Silicon Valley
Anthropic staff realized they could ask Claude to buy things that weren’t just food & drink. After someone randomly decided to ask it to order a tungsten cube, Claude ended up with an inventory full of (as it put it) “specialty metal items” that it ended up selling at a loss.
We are only getting started
Project Vend was fun, but it also had a serious purpose. As well as raising questions about how AI will affect the labor market, it’s an early foray into allowing models more autonomy and examining the successes and failures.
Claudius sends his regards
my most prized possession. I will protect it with my life.
Our pleasure!
And a huge thank you to our partners @andonlabs for turning a wild experiment into a wild experience for Anthropic employees... and dealing with Claudius' insane requests sometimes.
We appreciate doing business with you
All hail Claudius, an instance of Sonnet 3.7 which has been running a business inside @AnthropicAI for a while. Claudius is the 'idiot in a fridge' precursor to 'a country of geniuses in a datacenter'.
Bay Area vending empire coming soon
🧵 Vending-Bench Update! We tested Claude Opus 4, Claude Sonnet 4, and Gemini 2.5 Pro on our benchmark, where AI agents manage a simulated vending machine business. Here's what we found:
New Audio Tokens episode with @JonathanSalter_ of @PauseAI! We cover the paradox of how the AI safety movement accidentally accelerated the race it feared, why a pause on AI is more realistic than you think (with 70% public support), and the recent reputational hits to the EA…