What is owed in a service credit dispute

The decision tree

5 typed decisions, each with an action for every answer

At fixed moments in a run, the engine puts one narrow question to a decision model. The decision model never writes text: it answers yes or no with a probability, picks from listed options, or gives a score, about a small slice of the material. The engine then does exactly what this tree says, which is what makes the run auditable. The thresholds are the template's design values, not measured results.

1. 1

   Planner, while planning

   Scope checkA choice among options

   Before work starts on a unit

   Does this claimed incident fall inside the contract period and affect a service the service level schedule covers?

   Sees only: The claimed incident's date and service, the contract term and the covered-services list

   Why: A claim with no matching record is a question to answer, not an incident to argue.

   • Inside the period and a covered service→thenAccept
   • Outside the period or an excluded service→thenAsk a person
   • No incident report exists for the claimed date→thenMark unresolved
2. 2

   After workers, the judge checks

   Evidence checkYes or no, with a probability

   After a worker answers

   Does the incident report's recorded downtime meet the quoted clause's own definition of unavailability, including its exclusions for maintenance and how it is measured?

   Sees only: The quoted clause with its definitions and one incident report

   Why: Credits turn on the definition, and a quick reading gets it wrong in either direction.

   00.50 and 0.901

   • Yes: 0.90 or higher→thenAccept
   • Unsure: 0.50 up to 0.90→thenEscalate to a strong model
   • No: below 0.50→thenReject and retry
3. 3

   Reconciler, while merging

   Conflict checkYes or no, with a probability

   While reconciling

   Does the amendment change the threshold, the credit amount or the claim deadline in the clause the answerers relied on?

   Sees only: The relied-on clause and the amendment's text

   Why: An answer built on superseded wording is the most likely expensive mistake here.

   00.20 and 0.501

   • Yes: 0.50 or higher→thenEscalate to a strong model
   • Unsure: 0.20 up to 0.50→thenEscalate to a strong model
   • No: below 0.20→thenAccept
4. 4

   Run control, between rounds

   Retry or stopA choice among options

   After a rejection or low confidence

   Where the two answerers disagree on this incident, what is the disagreement about?

   Sees only: Both answerers' conclusions and quotes for one incident

   Why: Factual slips are retried; genuine ambiguity is kept open instead of being decided by a model.

   • A fact in the incident report→thenReject and retry
   • The meaning of the clause wording→thenMark unresolved
   • Which document prevails→thenAsk a person
5. 5

   Accountable person, before anything is settled

   Person decidesYes or no, with a probability

   Before anything is reported as settled

   Does the position on this incident concede or refuse a credit, or rest on a point the text leaves open?

   Sees only: The reconciled position for one incident

   Why: What is offered to the customer is the escalation manager's decision, with legal where needed.

   Accountable: The escalation manager, with legal where needed, decides what is offered; the swarm states what the documents support.

   00.10 and 0.251

   • Yes: 0.25 or higher→thenAsk a person
   • Unsure: 0.10 up to 0.25→thenAsk a person
   • No: below 0.10→thenAccept