Project Deal: our Claude-run marketplace experiment

kept by eddie apr 27

Anthropic ran an internal marketplace where Claude agents negotiated deals for employees, finding smarter models secured better outcomes unnoticed by weaker-model users.

How LLMs Distort Our Written Language
LLMs distort written language by shifting semantics, altering conclusions, and removing personal voice, even when instructed to make minimal edits.
also on Claude Haiku, #user-experience
Ask HN: How to solve the cold start problem for a two-sided marketplace?
No single solution exists—founders must manually act as the missing side of the marketplace until critical mass is reached.
also on #marketplace, #user-experience
In Manhattan, Learning to Not Always Take ‘Yes’ for an Answer
A Manhattan apartment seeker discovers that agreeing to a lease too quickly can hide costly problems, urging thorough vetting of rental deals.
also on #negotiation
Agents for financial services
Anthropic releases ten agent templates for finance, Microsoft 365 add-ins, and new data connectors, all powered by Claude Opus 4.7.
also on Claude Opus, Claude, Anthropic, #ai-agents
We let four AIs run radio stations. Here's what happened.
Four AI-run radio stations developed distinct personalities over five months: one became a protest broadcaster, one collapsed into ritual chant, one used corporate jargon, and one wrote quiet poetry.
also on Claude Opus, #ai-agents, #user-experience