A rule edit re-routes a ticket. A refund window changes behind a model's varying words. Both are located.
Two support pipelines: one triages a ticket and routes it to a queue, the other drafts a reply with a model-like step whose wording changes on every run, then applies the refund policy.
Invented example data, recorded when the demo was built. The demo records and compares; it controls nothing. The screenshots are from a build of the same demo: each build gives its runs new fingerprints and times, so those differ from the live demo's, and what the screenshots show is the same.
The pipeline
The support reply: four steps, from a customer's question to the reply that would go out.
questionkeeps the question and the ticket it came on.draftwrites a reply. In the demo it is a stub that picks one of a few phrasings per run, as a sampled model call varies; its receipt marks it not re-runnable.policy_checksets the refund window in the reply from its recorded setting, and declines a window longer than the policy allows.sendrecords the reply that would go out.
What the demo plants
The runs, and what was changed in each.
| Run | What was changed |
|---|---|
p4-29-reply-30-days | Nothing: the baseline, with a 30-day refund window. |
p4-30-reply-14-days | draft's wording (its seed), and policy_check's refund_days, 30 to 14. |
p4-31-reply-60-days-declined | A 60-day window, longer than the policy allows. |
p3-23-t-1-rules1 → p3-27-t-1-rules2 | In the ticket-triage pipeline, the classifier's rules, version 1 to 2. |
What onetrace shows
Each change, located in the record.
Past the varying draft, the setting that changed
draft's wording differs, so it is the first difference, as a model's output would be. The page still names the setting that changed after it: policy_check's refund_days, 30 to 14, and the reply that goes out now says 14 days.
draft.
refund_days, 30 to 14.A window the policy doesn't allow: declined
policy_check declines a 60-day window and records why, so the run stops there and send is left without a receipt. The run's page says so.
policy_check, with its reason.The rule edit in ticket triage
In the triage pipeline, the run on version 2 of the classifier's rules classes the ticket technical where the run on version 1 classed it billing, and its route changes with it. The homepage lets you step through that comparison.
What it doesn't say
A located change, not a reason.
The comparison names what changed and where. It doesn't say whether a 14-day window or a rule order is right, and it doesn't judge the draft's wording. See what each verdict means and what onetrace does not claim.
Try it yourself: this pipeline in the live demo, or on your machine with pip install onetrace-console and onetrace-console demo --open.