Add to your use case

You're updating We ran an agent swarm (34 to 56 agents) on Claude Opus 5.5 against live Polymarket markets, locked the forecasts before resolution, then scored them. The swarm was one Opus call at 500x the price. Tell us what changed, results since, or what you'd do differently. We review every update before it goes on the card.