How does escalation connect to the other records in a lending process?
Escalation is a set of links between application, authority, referral, note, decision and task records, and each link carries state changes that an agent or grader can follow.
In a generated lending environment, escalation is a link between records rather than a standalone table. A useful release makes those links explicit so a model or agent can follow them.
Typical relationships:
- `application` (`APP-2187`) holds requested amount, product and current status - `authority_matrix` holds role, product and maximum exposure per decision type - `referral` (`REF-0931`) points to `application_id` and carries `from_role`, `to_role`, `reason_code` and `created_at` - `handover_note` points to `referral_id` and carries recommendation text and an evidence list - `decision` points to `referral_id` with outcome, conditions and the deciding role - `task` records point back to the application and are created by the decision
The key integration detail is **which record changes state when**. Creating a referral should move the application to an escalated status and remove the decision task from the original reviewer. A decision record should close the referral and, if conditions exist, create tasks.
For linked facilities, an `exposure_group` table lets combined exposure trigger the referral even when each facility is under the limit. Ask for that join if you want to test whether a model aggregates before checking authority.
Formats are typically JSONL per table with stable identifiers, and Parquet for larger tables. Confirm identifier style and foreign-key coverage in a sample before scoping further.
Explicit links and state changes let a model or grader trace an escalation from trigger to follow-up task.
- Audience:
- Model labs
- Content type:
- Integration
- Buyer question:
- What it is and why it matters
Content release: r-a220eb6e099b243b
Related questions
- Why test escalation separately from approve and decline decisions?
- What does a real escalation look like when the trigger sits in a related record?
- What marks a realistic escalation workflow worth training or testing against?
- What tool actions and linked records does an escalation agent work with?
- Why does escalation matter when building agents for credit decisions?
- How does an escalation map onto an environment's state machine and tool actions?