Input
What you provide
- Baseline and candidate versions with the cases you kept.
- Same cases, same rubric, same thresholds.
AuraOne / Products
Loading the product.
Loading capabilities, workflow, and program details.
AuraOne / Products
Loading capabilities, workflow, and program details.
Models / Regression Bank
Lock the comparison context. Inspect what changed. Keep every new failure with its owner.
Evaluation teams comparing candidates against retained failure cases. Each program is configured around the workflow, review policy, and evidence your team needs.
Input
Work
Output
Move from intake to review and handoff with clear owners at every step.
01
Baseline and candidate versions with the cases you kept.
Locked comparison
02
Replay. Compare the deltas. Retain the new failures.
Failure deltas
03
Same cases, same rubric, same thresholds. The evaluation owner and a domain reviewer.
Named review and decision history
04
A regression verdict with failed cases and recommendation.
Retained cases
What you receive
| Outcome | Work | What you receive | Program fit |
|---|---|---|---|
| Advance | The required Regression Bank checks and review are complete. | Locked comparison, Failure deltas, Retained cases | Program fit depends on the source and its version. It also depends on criteria, reviewers, and operating scope. |
| Hold | A required check, reviewer, or approval is missing. Or a source record is unresolved. | What is blocking, who owns it, and the recovery it needs. | The responsible owner and recovery step stay visible until the issue is resolved. |
| Return for work | The source object or workflow requires correction and another review. | The returned items, their owner, and the review route back. | The updated work returns through the same review path before a new decision. |
Inspect the workflow
AuraOne / Models / Regression Bank
Lock the comparison context. Inspect what changed. Keep every new failure with its owner.
Evaluation teams comparing candidates against retained failure cases.
Input
Work
Output
Move from intake to review and handoff with clear owners at every step.
01
Baseline and candidate versions with the cases you kept.
Locked comparison
02
Replay. Compare the deltas. Retain the new failures.
Failure deltas
03
Same cases, same rubric, same thresholds.
The evaluation owner and a domain reviewer.: Select the candidate, hold it, or bank a new case.
04
A regression verdict with failed cases and recommendation.
Retained cases
What you receive
| Outcome | Work | What you receive | Program fit |
|---|---|---|---|
| Scoped workflow | Replay. Compare the deltas. Retain the new failures. | Locked comparison | Configured to your workflow and goals. |
| Review decision | Select the candidate, hold it, or bank a new case. | Failure deltas | Configured to your workflow and goals. |
| Retained handoff | A regression verdict with failed cases and recommendation. | Retained cases | Configured to your workflow and goals. |
Next step
Bring your source material and its owners. Bring the acceptance criteria and any operating constraints. We will turn them into a scoped plan.
Aggregate numbers publish automatically once enough organizations contribute to the bank. Until then we keep the story, not the scoreboard.
Regression Bank turns escapes into replayable tests. When configured, it can gate deployments on program-defined thresholds and publish evidence into your dashboards.
Guardrail replay
Regression Bank