AICOMPASS — Agent Governance

Registered Agents16▲ 3 this month

Action Permissions9Requires controls

Awaiting Review52 security reviews

Approved Agents11Ready for use

Kill Switch Ready1487% coverage

High-Risk Actions4Needs oversight

Agent Register

AI agents with autonomy level, tool access, controls, and review state.

View agent register

Agent	Use Case	Autonomy	Tool Access	Risk	Kill Switch	Status	Owner
FR Finance Reconciliation AgentAGT-2025-001	Finance Reconciliation	Act with Approval	ERP, Email, Jira	High	Ready	Security Review	Nina Gupta
CS Customer Support AgentAGT-2025-002	Customer Support Assistant	Recommend	CRM, Knowledge Base	Medium	Ready	Approved	Riley Morgan
HR HR Screening AgentAGT-2025-003	HR Screening Assistant	Recommend	ATS, Document Store	High	Ready	High-Risk Review	Jamie Singh
LD Legal Drafting AgentAGT-2025-004	Legal Drafting Copilot	Assistive	DMS, Clause Library	Medium	Pending	Pending Review	Avery Taylor
CP Claims Triage AgentAGT-2025-005	Claims Triage Assistant	Act with Approval	Claims System	Medium	Ready	Approved	Morgan Kim
PR AI Pricing AgentAGT-2025-006	AI Pricing Engine	Act Automatically	Pricing API, CRM	Critical	Missing	Blocked	Laura Müller

Finance Reconciliation Agent

Selected agent — AGT-2025-001

High Risk

Governance Rationale

This agent can prepare reconciliation actions and initiate workflow updates. All write actions require approval, logging, and reversible execution.

AutonomyAct with Approval

Risk LevelHigh

Tool AccessERP, Email, Jira

Kill SwitchReady

ReviewerSecurity

Next ReviewJun 12, 2025

Permission Matrix

Read ERP reconciliation recordsAllowed without approval

Allowed

Draft reconciliation adjustmentRequires finance reviewer

Approval

Send customer/vendor emailHuman approval required

Approval

Post journal entryNot allowed for this agent

Blocked

Risk Events

Recent permission, tool-use, and escalation events.

Review event log

Event	Agent	Action	Severity	Decision	Time
Approval required	Finance Reconciliation Agent	Draft adjustment	High	Pending	10:18 AM
Blocked action	AI Pricing Agent	Auto-price update	Critical	Blocked	09:45 AM
Escalation triggered	HR Screening Agent	Candidate ranking	High	Review	Yesterday
Tool access changed	Legal Drafting Agent	DMS read access	Medium	Approved	Yesterday

Autonomy Distribution

Agents by action authority.

Assistive

Recommend

Act w/ Approval

Act Auto

Required Controls

Controls required before agent approval.

Review control gaps

Permission matrix approvedPending

Kill switch testedPassed

Action logs enabledEnabled

Human approval workflowIn Review

Escalation route configuredDone

Incident playbook linkedMissing