How Mission Policy coordinates multiple AI agents with shared budgets, phases, and constraints.
4 agents plan a $5,000 trip to Barcelona. Budget flows dynamically — cheap flights mean a nicer hotel.
Flights found a deal at $800. Hotel now has access to $2,700 instead of a fixed $1,750.
Flights: $800. Hotel: $1,900. Activities: $2,300. Total: $5,000. Entertainment budget doubled because the flight agent found a deal.
Production is down. AI agents diagnose and fix — with increasing authority and budget at each escalation level.
Diagnosed in 40 min ($120). Fixed by scaling up ($2,100). Total: $2,220 of $10,000 budget. Escalation L3 never activated.
3 agents test different ad channels. Small seed budget each. The best performer wins the full campaign budget.
Facebook won at $8 CPA. Scaled to $5,000 → 625 conversions. Google and Twitter stopped automatically. Total spent: $5,600.
Buy 50 laptops. Agents search suppliers, compliance checks the winner, only then purchase is authorized.
If compliance fails → rollback to search, exclude failed vendor
Supplier B: 50 Lenovo ThinkPads at $1,350 ($67,500). Compliance passed. Purchased. $7,500 under budget.
One support agent, three authority levels. Starts with discount codes, escalates to full refunds if needed.
Resolved at L2 for $85 of $500 max budget. L3 never reached. Customer satisfied.
Read the full specification with all objects, fields, and evaluation rules.