Assess Trust And Control Risks
assess_trust_control_risksRead-onlyIdempotent
Reviews proposed AI workflows for trust and control risks by analyzing human oversight, data boundaries, quality controls, escalation triggers, and automation safety.
Instructions
Reviews a proposed AI workflow for human review needs, data boundaries, quality controls, escalation triggers, unsafe automation risks, and confidence.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| aiIdea | No | Specific AI or automation idea the user is considering. | |
| businessType | No | Type of business being reviewed. | |
| riskConcerns | No | Known concerns about customer trust, brand risk, compliance, money, privacy, or quality control. | |
| workflowIdea | No | AI workflow or agent idea to assess for trust and control risk. | |
| canAffectMoney | No | Whether the workflow can influence pricing, refunds, payments, scope, or financial decisions. | |
| currentProblem | No | Main business or workflow problem the user wants to solve. | |
| customerFacing | No | Whether the workflow directly affects customers or customer-facing messages. | |
| currentControls | No | Existing review, approval, QA, escalation, or monitoring controls. | |
| currentWorkflow | No | Current workflow before AI or automation. | |
| proposedWorkflow | No | Proposed workflow design, including what AI would do and what humans would review. | |
| usesSensitiveData | No | Whether the workflow may use private, regulated, or sensitive business/customer data. | |
| canAffectBrandTrust | No | Whether the workflow can affect brand reputation, customer expectations, or public trust. | |
| requiresExpertJudgment | No | Whether the workflow needs professional, strategic, legal, financial, HR, or domain judgment. |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| riskLevel | Yes | Estimated risk level for the workflow or recommendation. | |
| confidence | Yes | Confidence level based on the clarity and completeness of the provided business context. | |
| riskSummary | Yes | Concise public-safe summary of the main trust and control risks. | |
| humanReviewRules | Yes | Rules for when a human must review, approve, or override AI output. | |
| requiredControls | Yes | Controls needed before the workflow should be piloted or expanded. | |
| escalationTriggers | Yes | Conditions that should escalate to a human owner or expert reviewer. | |
| missingInformation | Yes | Information the user should provide to improve risk assessment confidence. | |
| dataBoundaryWarnings | Yes | Warnings about sensitive data, privacy, access boundaries, or inappropriate inputs. | |
| notRecommendedActions | Yes | Actions that should not be automated in the current version. |