Can you learn an evolving agent router just by simulating an auction?
No outcome rewards, no pre-trained weights, just a bidding war on the best plan and some shared memory.
@sxbrinahhu led last week's NYC AI reading group where we read Scaling Small Agents Through Strategy
