AI Connections
How it works: 16 terms from this Drop's news make 4 groups of 4. You can make 4 mistakes. About 3 minutes. Your streak lives only on this device.
Mistakes left: 4
Show the groups
Anthropic's IPO, by the numbers
$895.4 billion · 40 of 41 · $1.5 trillion · prospectus
CNBC reports Anthropic's amended prospectus lists about $895.4 billion in commitments, voters opposed data center measures in 40 of 41 cases, and investors see up to $1.5 trillion.
From the Drop: Anthropic lists $895.4 billion in commitments as voters sour on data centers
AWS's remote Claude Code sample
microVM · 8 hours · three attempts · never merges
Each task gets its own AgentCore microVM and a GitHub token that expires after 8 hours, has up to three attempts, and the agent opens pull requests but never merges them.
From the Drop: AWS sample runs Claude Code as a remote agent that opens PRs but never merges
Taxbit's compliance agent
55 percent · 24 hours · support tickets · tax data
Taxbit says more than 55 percent of askers return within 24 hours; the agent has no access to customer tax data and can only write the caller's own support tickets.
From the Drop: Taxbit's compliance agent never lets the model decide whose permissions to check
Anthropic's EVE experiments
saboteur · EVE Frontier · four agents · Fellows
In an Anthropic Fellows project, teams of four Claude agents in EVE Frontier rarely spotted a teammate secretly told to undermine them.
From the Drop: Anthropic: Claude agent teams in EVE Frontier almost never caught a secret saboteur
Games are written by AI from this Drop's sourced items. Scenarios are illustrative.