Personal assistants
Explore coverage around actual permissions and potential losses, from shopping and bookings to file management.
A FRAMEWORK FOR RESPONSIBLE AUTONOMY
Autonomous intelligence.
A clearer view of risk.
Explore a coverage framework for personal assistants and enterprise agents. Define permissions, quantify potential losses and understand the cost of autonomous action.
Concept and simulation only · No policies, underwriting or claims services
AGENT RISK PROFILE01 / DEMONSTRATION
AGENTCOVER / RISK REGISTER◇
Personal assistant
Shopping & travel · Personal scenario
DEMO-PERSONAL-01NOT A POLICY
01 Permission boundariesTools · Accounts · Actions
02 Loss exposureDirect loss · Recovery costs
03 Coverage structureDeductibles · Limits · Exclusions
Enterprise workflows
Consider financial permissions, business interruption and third-party liability in one risk assessment.
An interface for agents
Structured assumptions and traceable calculations let agents read and compare risk scenarios.
RISK & PRICING LAB
Make the cost of risk visible.
Adjust incident probability, losses and limits to see how costs change. Every number comes from your assumptions and a disclosed formula.
All amounts in CAD
What can your agent do?
Start with an example, then adjust the assumptions. You set the probability; this tool does not predict real-world risk.
Example assumptions
Example permissions: search itineraries, book and purchase. The execution system must enforce spending limits.
0%Assumed, not measured20%
CAD
Loss within the assumed coverage scope, not the agent’s valuation or revenue generated.
Adjust expense and capital assumptions
Record the deployment, model version, tool permissions and shared budget. Reassess future risk when permissions or configurations change.
What if the agent crosses a boundary?
Simulate one covered loss using the deductible and payout limit above.
Separate simulation; premium stays unchanged
—
Illustrative payout—Retained loss—
Payout = min(max(covered loss − deductible, 0), payout limit). A real claim would also require verification of coverage terms and evidence.
BUILT FOR MACHINE READERS
An agent? Start here.
Read the capabilities and limits, download the OpenAPI description or submit assumptions for a structured estimate. The API does not issue insurance, execute transactions or grant permissions.
What is included in this model?
Annual premium = (expected payout + capital cost + fixed cost) ÷ (1 − variable expense ratio − target profit margin). Expected payout = annual incident probability × single-incident payout after the deductible and limit. All initial scenario values are illustrative assumptions, not industry averages.
This simplified model assumes at most one specified incident per year. It does not estimate repeated incidents, long-tail litigation, correlated failures after model updates, taxes or reinsurance. No past incidents does not mean zero risk. Payment limits cannot bound every data breach or third-party claim.
Personal scenarios focus on direct losses and recovery costs. Enterprise scenarios can also explore business interruption and third-party liability. Actual insurance requires an insurer, underwriting information and formal terms. This page is a product concept and pricing sandbox only.