Feature

Audit & Control

AI moves fast. Averence keeps humans in the loop — with a complete audit trail of every agent interaction, configurable approval gates, and real-time oversight of what's happening on your platform.

Why It Matters

Autonomous Agents Need Human Guardrails

AI agents can act at machine speed. Without controls in place, an agent might call sensitive tools, trigger high-value actions, or access information in ways that weren't intended. Averence's Audit & Control layer intercepts these situations — giving you the visibility and tools to respond.

This isn't about blocking agents. It's about ensuring your business operates within safe, auditable boundaries while still being fully open to the agentic economy.

Governance Rules

Configurable Approval Gates

Define rules that pause agent execution and notify a human reviewer before high-stakes actions proceed. Rules are evaluated in priority order for every tool call:

Rule: Require approval for demo booking

If book_demo is called by an unverified agent → pause, notify reviewer via webhook before confirming.

ACTION: PAUSE + NOTIFY

Rule: Allow unrestricted information queries

get_pricing, get_product_detail, and answer_faq from any grade → allow with no approval required.

ACTION: ALLOW

Rule: Block repeated Act calls from unverified bots

More than 3 Act-type tool calls from a Grade C or F session within 10 minutes → block and flag for review.

ACTION: BLOCK + FLAG

Rules are expressed as JSON and can be updated via API without a deployment:

{
  "rule_id": "demo-booking-gate",
  "priority": 10,
  "trigger": {
    "tool": "book_demo",
    "condition": "agent_grade != 'A'"
  },
  "action": "pause",
  "notification": {
    "webhook": "https://your-business.com/webhooks/portiko",
    "timeout_seconds": 300,
    "on_timeout": "reject"
  }
}

Audit Trail

Every Interaction on the Record

Every tool call, governance rule evaluation, and decision is written to an immutable audit log — queryable in the Partner Dashboard and exportable for compliance review:

{
  "timestamp": "2026-04-10T06:44:01Z",
  "session_id": "sess_abc123",
  "agent_grade": "B",
  "tool": "get_pricing",
  "input": { "plan": "enterprise" },
  "governance_rule_triggered": null,
  "decision": "allowed",
  "response_latency_ms": 84
}

What You Can Monitor

Partner Dashboard

Live Agent Activity Stream — Real-time feed of all tool calls, grouped by session
Traffic Grade Distribution — Breakdown of A/B/C/F traffic hitting your platform
Read vs. Act Split — Volume of information queries vs. action calls over time
Governance Events — Pending approvals, blocked actions, flagged sessions
Tool Call Volume — Which tools are being called most, and by which agent grades

SOC 2 Auditable GDPR Ready Role-Based Access Human-in-the-Loop

Request Partner Access → ← Traffic Governance