When a plan stops working, the decision that follows either contains the damage or creates more of it.
ACE is a decision framework built from the operational reality of managing high-stakes service itineraries — where a single change can cascade across multiple dependencies in minutes. It structures the judgment moment between pressure and action, making decisions explicit, reviewable, and accountable before anything reaches the client.
In high-stakes service operations, the most dangerous moment is the one between a problem appearing and an action being taken.
ACE was built from the experience of managing complex travel itineraries for high-value clients — where a flight delay, a supplier cancellation, or a missed window can cascade across transfers, hotel check-ins, dinner reservations, and client communications in minutes. The problem is never the disruption itself. It is the unstructured decision that follows it.
A system that fails gracefully is more trustworthy than one that always produces an answer. ACE does not automate judgment — it structures it. The framework applies wherever a professional must make a consequential decision under time pressure with incomplete information: travel advisory, client experience operations, onboarding coordination, or any service where the cost of a wrong decision is borne by someone who trusted you with their outcome.
One operating framework.
Five safeguard layers.
ACE separates operational judgment into three stages — Situation, Decision, Confirmation — with risk classification, contingency gating, human review, and decision logging running as enforcement layers beneath.
Decision Taxonomy
Each decision option has a defined risk profile, operational dependency, required confirmation, and contingency requirement. These are not suggestions — they are the system's decision model.
| Decision Option | When Used | Risk | Required Confirmation | Contingency |
|---|---|---|---|---|
| Rebook / reroute | Dependency at risk, viable alternative exists | High | Supplier confirmed, client aware | Fallback path if rebook fails |
| Hold and monitor | Situation unstable, window still open | Medium | Update threshold defined | Escalation trigger if window closes |
| Proactive client communication | Client impact likely, time to act exists | Medium | Message reviewed before sending | Recovery offer if needed |
| Escalate | Risk exceeds advisor control | High | Senior review completed | Manual recovery path |
| Manual recovery | Guided path not supported by system | Critical | Override reason required | Full audit log entry |
Risk Classification Tree
When a decision is high risk, ACE requires a contingency path before confirmation can proceed. If no safe recommendation exists, the system surfaces manual recovery or escalation — it does not guess.
Blocking Rules
The system disables confirmation when required conditions are missing. These are not warnings — confirmation is structurally blocked until the condition is met.
| Condition | System Behavior |
|---|---|
| High risk + no contingency | Confirmation disabled |
| Missing readiness check | Decision cannot close |
| Human review not completed | Action blocked |
| Required log fields incomplete | Decision remains open |
| No safe recommendation available | Manual recovery or escalation required |
The Decision Log
Every confirmed decision creates an immutable record. This is not a notes field — it is the accountability artifact that supports service recovery investigations and manager review after the moment has passed.
In manual recovery and escalation modes, all override actions are logged with elevated severity. A reason is required for every action taken outside the guided path.
The prototype in motion — two scenarios, one system.
The screens below follow two live cases through the full ACE flow: Luciana Ferretti (Decision Required — viable path found) and Marcus Webb (Critical Intervention — guided path not supported). Together they show the full decision range the system handles.
Today's Arrival Queue · Apr 20 2026 · Prioritized by failure risk · 1 Critical · 1 Decision Required · 2 In Progress · 1 Stable
Situation · Flight BA 447 delayed 83 minutes · Transfer window now 12m — below 25-minute minimum · Affected: Transfer, Hotel check-in, Client communication
Decision · AI drafted recommendation · Confidence: High · Rebook private transfer to 19:15 pickup · Option expires in 18m · Alternative paths available
Confirmation · You are about to: Rebook private transfer to 19:15 pickup · Affected: Transfer, Hotel check-in, Client communication · Cost: +£0 · Executed by: M. Okafor · Cannot be undone without manual recovery
Execution · 3 of 4 steps complete · Driver confirmation: pending → blocked
Execution blocked · Driver confirmation cannot be completed · Case cannot be marked Resolved until block is cleared or classified as Partially Resolved
Resolution · Partially Resolved · Client notification still pending · Archive blocked until communication dependency is cleared
Situation · Primary transfer supplier cancelled. No backup on record. Client dinner at 19:30 — missed reservation and stranded arrival at risk.
Decision · No safe recommendation available · Conditions beyond system decision model · Manual recovery or escalation required
Manual Recovery · All override actions logged with elevated severity · Override Action required · Reason for Override required · Cannot submit without both fields
Case Escalated · Ownership transferred to escalation team · Full handoff record written to audit log · Navigation returns to queue
Situation · Arrival Plan Stable · Everything is aligned. No action required. Monitoring continues in background. · EK 007 · DXB T3 · Atlantis The Palm · ETA 19:20
Measured by decision quality, not speed alone.
Faster decisions only count as success if rework and escalation do not increase. The primary metric is the Validated Decision Rate — decisions completed with the required situation, risk, contingency, and confirmation before action.
Measures whether decisions are completed with the required situation, risk, contingency, and confirmation before action.
Measures whether structured decisions actually reduce rework. Faster decisions that produce more rework are a failure.
Measures whether the decision structure is actually reducing post-action escalation — the primary cost of the original problem.
High-risk decisions with contingency defined before execution.
High-risk decisions that completed the human review gate before executing.
Confirmed decisions with all required log fields completed at time of action.
What ACE is
What ACE is not
The scope was defined to prove whether structured decision logic improves clarity before expanding into automation or system integrations.
ACE is not a task manager. It is a decision model.
It applies wherever a professional must make a consequential decision under time pressure, with incomplete information, where the cost of the wrong choice is borne by someone who trusted them with their outcome. Travel advisory. Client experience operations. Onboarding coordination. High-stakes service management.
The framework was built from practice — from the real operational experience of managing complex itineraries where a single delay cascades into multiple failures, and where individual judgment without structure is the system's weakest point.
Together with Aethel, it proves a consistent design thesis: service failures are usually legibility failures. Aethel makes progress visible to the client. ACE makes the decision visible to the advisor. Both are the same problem — solved at different points in the service chain.
Experience the decision flow live.
Open any case, then use the Simulate controls — Escalate, Advance Time, Destabilize — to trigger different system states and see how ACE responds to changing conditions.