Run hundreds.
Control every dollar.
Procurement, support, research, market analysis. Run a fleet of autonomous agents under one budget. See every agent and every payment in one view.
One bad agent drains a budget.
200 procurement agents negotiate across 15 vendors. Each needs a wallet, caps, and rules. Without a central guardrail, one misconfigured agent empties the budget in seconds.
A finance director who reviews every transaction manually cannot keep up beyond a handful of agents. The review backlog grows. Anomalies surface days after the spend already happened. The compliance team cannot prove that every payment was policy-checked.
The fleet problem is a control problem. The control needs to live inside the policy engine, not inside a human inbox.
One view. Every payment.
The control lives inside the protocol. Three primitives cover the fleets that procurement, research, and support teams actually run.
Hierarchical policies
Define organization-level spending limits. Set department-level budgets within those limits. Assign per-agent caps within department budgets. The hierarchy ensures that no individual agent can exceed its allocation, and no department can exceed its budget.
Real-time fleet dashboard
The Oris command center displays every active agent, its current spending, its KYA level, and its policy status. Anomalies surface immediately. Agents that deviate from their behavioral baseline trigger automatic alerts.
Automated suspension
When an agent exceeds its policy limits or exhibits behavioral drift, Oris suspends the agent automatically. The suspension lands inside the policy evaluation cycle. No human intervention is required for the initial response.
Org. Department. Agent.
Each level enforces its own cap. A department cannot exceed the org. An agent cannot exceed the department. The math holds end to end.
The organization sets the monthly cap. Each department draws against the organization envelope. Each agent draws against its department envelope. The evaluator checks all three counters atomically on every transaction.
A breach at any level blocks the transaction at that level. The activity log records which level rejected the spend and the exact counter state at evaluation time.
Spin up a fleet in one flow.
The SDK creates the fleet, attaches the budget, and binds each agent to its policy. Wallets, rules, and monitoring spin up together.
Each agent inherits its department policy and the organization envelope. The next transaction every agent attempts is evaluated against all three layers. There is no separate deploy or migration step.
# Create a fleet with hierarchical budgets fleet = oris.fleets.create( name="procurement-fleet", monthly_budget=50000.00, ) # Add agents to the fleet for vendor in vendor_list: agent = oris.agents.create( fleet_id=fleet.id, name=f"buyer-{vendor.name}", policy={ "max_daily": 500.00, "max_per_tx": 100.00, "allowed_categories": ["compute"], }, )
Fleets in three industries.
The shape varies. The controls stay the same. Three sectors already run multi-agent fleets under cascading budgets.
Enterprise IT
Hundreds of agents purchasing SaaS licenses, cloud compute, and infrastructure services across multiple vendors. Centralized budgets prevent overspending. The activity log provides a complete audit trail for finance.
Financial services
Agent fleets executing market research, data acquisition, and counterparty analysis with strict per-agent spending controls. Compliance teams receive real-time visibility into every transaction.
Supply chain
Procurement agents negotiating with suppliers, comparing prices, and executing purchase orders within approved budgets. Each agent operates within its assigned vendor category and spending limits.
Run a fleet this week.
Sign up, set your policies, deploy agents in minutes. Wallets, caps, and the log all spin up together.