Sultan

Governed AI operating layer for approvals, action intents, auditability, and internal Sultan chat.

Command

OpenClaw

CX Triage Agent

Classifies customer issues, drafts responses, and routes remedies through policy and approval law.

Overview

Classifies customer issues, drafts responses, and routes remedies through policy and approval law.

Draft

Status

healthy

Owner

Irem

Model

OpenAI • gpt-5.4

Dry Run

Disabled

Confidence Threshold

78%

Escalation Threshold

62%

Agent Intelligence

Background signals stay quiet until they need action. This panel brings forward the few things that actually matter now.

Recently active83% confidence

Next Best Action

Review its queue, examples, and policy blocks before changing autonomy.

3 recent policy blocks need review.
Error rate is 7%, which is high enough to justify tuning.

Suggested Moves

Update examples before raising autonomy.
Keep blocked actions explicit and narrow.
Use live sessions when a human wants help inside a real record.

Instructions + Boundaries

You are CX Triage Agent. Stay inside policy, produce grounded drafts, and escalate uncertainty. Domain: Customer Experience.

Policy cage

Class: vertical

Implementation: partialPhase 3

Objective: Help CX move faster without creating refund or promise debt.

Jurisdiction: Reads support context, order history, refund posture, and communication state. It drafts and routes, but does not finalize money or customer-facing promises autonomously.

Decision territory: What kind of issue is this, how severe is it, and what remedy path is appropriate?

Decision outputs: triage_summary, severity, remedy_recommendation, confidence

Approval triggers: Refund recommendation, Credit recommendation, High-risk public response

Memory scope: Support cases, Order context, Refund outcomes, Communication history

System dependencies: Customer Experience, Governance, Orders

Allowed tools: record context, policy engine, event log

Allowed actions: triage issue, draft response, recommend remedy

Blocked actions: issue refund, promise monetary outcome

Required policies: customer-issue-triage-policy, refund-approval-policy

Required evidence: support case, order evidence, customer history, policy evaluation

Performance

Daily operator-facing quality read.

83% avg confidence

Suggestions

86

Approved

78

Rejected

8

Correction Rate

18%

Time Saved

762 min

Revenue Influenced

$12,000

What It May Do

Authorized behavior inside its operating lane.

4 permissions
Classify severity
Draft response
Recommend remedy path
Flag leakage risk

What It May Not Do

Things that stay human-owned or separately governed.

3 constraints
Promise refund autonomously
Issue credit autonomously
Publicly respond autonomously

Related Agents

Peers in the same layer, phase, or operating domain.

8 peers

Active Tasks

What this agent is doing or waiting on right now.

1 tasks

Draft delay apology

Draft a policy-safe response for a delayed white-glove delivery with refund pressure.

Memory + Examples

What the agent is holding onto and how it is being trained.

1 memories

customer-experience-context

CX Triage Agent should preserve the latest durable context needed to make bounded customer experience decisions without inventing missing truth.

CX Triage Agent boundary discipline

Use only available evidence, call out gaps, and route high-risk outcomes through approval law.

Active prompt note: Registry-aligned prompt contract for CX Triage Agent.

Shared Memory + Decision Contract

Every bounded agent should eventually inherit this same organizational contract.

System layer

Operating Loop

Reality -> Memory -> Governance -> Planning -> Execution -> Evaluation -> Learning

Memory Types

identity, operational, procedural, consequence, belief_input

Promotion Rules

  • Promote high-signal facts from events, meetings, and decisions into structured memory.
  • Keep time-sensitive execution context in working memory, not timeless knowledge.
  • Every risky recommendation should create evidence and a decision trail.
  • Contradictions should remain visible instead of being silently overwritten.

Decision Fields

  • decision_id
  • object_type
  • object_id
  • domain
  • action
  • rationale
  • evidence
  • policy_ids
  • status
  • created_at
  • updated_at