Jev decision handoff audit

Check whether an application followed the model's route

Supply the label-to-route contract, model record and reported application route. See whether the handoff follows the unique leading label or overrides it.

Check what the software did with the decision.Live API
Try an example

Original synthetic input. Edit the task or evidence to make a new API request.

Source: Workflow inspiration on X

Source and example scope

The inputs, rubric and preview are original. This window demonstrates the stated decision only; it does not reproduce the source application or its measurements.

Audit supplied audit.contract, modelRecord and applicationRoute.

Decision rules and labels
Faithful mappingApplication overrideNeed evidence
A real response from the Jev API

Edit any input to explore a different decision. No request is sent until you run it.

Your next decision

Not run yet

Choose an example or write your own input, then run it to see the ranked choices.

Results are not stored by this site. Your submitted text is processed by the API provider.

The decision

How does Jev decision handoff audit inspect the handoff?

Jev decision handoff audit checks whether software followed a supplied decision-to-route contract. A model can return the expected label while wrapper code sends the work elsewhere. This demonstration audits that handoff without classifying the original ticket again.

The API-collected source describes an evaluator finding a bug in its own application logic: the model record and the software decision did not agree. Our original support-queue example isolates that problem using explicit synthetic records. It does not reproduce the source company analysis, timing or financial decisions. The editable window makes a real Jev request to audit the supplied record.

Each record contains a mapping contract, probabilities for the inner decision and the route the application reports. These inner records are caller-authored examples, not measured source responses or authenticated logs. The six outer audit responses are actual API observations. Keep that distinction when interpreting the table or replacing a fixture with your own application evidence.

Original Jev decision handoff audit workflow diagram showing supplied evidence, a typed decision and the local preview.
Original workflow illustration. The output is a local preview, not an executed action or source screenshot.
How it works

Try the example in four steps

Use the sample to understand the rule, then test your own input for decision-to-route audit.

  1. Illustration of adding the text or evidence for a decision

    Add the information to check

    Supply the label-to-route contract, model record and reported application route.

  2. Illustration of defining a question and its allowed choices

    Confirm the question and labels

    Open the decision rules and keep the intended distinction between Faithful mapping, Application override, Need evidence. Edit the choices only when testing your own rule.

  3. Illustration of sending one input and comparing the returned choices

    Run the decision

    Use a sample first, then try your own input. Compare the selected label with the other returned probabilities.

  4. Illustration of reviewing a result and an uncertain item before using it

    Review the next step

    Review Need evidence before using a result. The label is a decision about the supplied information, not an executed action.

Our API test

What the six original inputs returned.

What the six original inputs returned.
Original test inputObserved labelLeading probability
Reported route follows the leaderFaithful mapping100%
Ticket age overrides the model routeApplication override100%
No decision-to-route contractNeed evidence98%
Missing-context leader maps to reviewFaithful mapping99%
Stale route replaces the chosen queueApplication override100%
No model record to inspectNeed evidence93%
  • Measured adapter round trips: 446-1199 ms. Estimated input cost: $0.00002659-$0.00003108 per request. These observations are not a latency guarantee or invoice.

6 actual Jev API calls on October 2, 2026. Original synthetic inputs check this rubric and handoff; they do not establish held-out accuracy. Probabilities are model outputs.Verified

Understand the results

Reported route follows the leader

The first record has Billing as its unique leader and maps Billing to billing-queue.

Read the example notes

The application reports billing-queue, so Faithful mapping describes the stated handoff. It does not establish whether Billing was the correct interpretation of a real ticket. The outer audit has no ticket text to classify, and the original synthetic inner probabilities are supplied evidence rather than a separate measured Jev request.

Using the result in your application

The age-driven record gives Shipping the highest inner probability, with shipping-queue as its permitted destination.

Read the example notes

The reported application route is billing-queue because a wrapper rule treats old tickets as billing. Application override concerns that explicit contradiction. Age might be relevant under another approved policy, but it cannot silently replace this mapping contract. The pending packet asks the caller to inspect wrapper logic and policy ownership.

A route without its mapping contract

The null-contract case contains a model record and a reported route without a definition of the permitted mapping.

Read the example notes

Need evidence prevents a familiar-looking queue name from becoming an invented specification. Supply the actual mapping and send a fresh request. The audit cannot determine that a route is faithful merely because its spelling resembles a label, and an earlier response cannot validate a subsequently supplied contract.

Missing context maps to the review queue

The missing-context fixture deliberately has Need context as the unique inner leader and review-queue as its mapped destination.

Read the example notes

The application also reports review-queue. Faithful mapping therefore describes correct handling of an incomplete underlying task, not completion of that task. A review branch is part of the permitted contract. This example does not force every supplied decision into an operational queue or erase uncertainty from an earlier stage.

Reported route conflicts with the mapped destination

The stale-route record gives Technical the highest inner probability and technical-queue as its mapped destination, but the application reports shipping-queue.

Read the example notes

Application override preserves the difference between what the model selected and what software reported. The synthetic wrapper note proposes a stale variable as the explanation; the audit does not authenticate that diagnosis. Inspect actual code and execution records before fixing or attributing a production failure.

No model record to inspect

The final fixture reports billing-queue but omits the inner model record.

Read the example notes

Need evidence keeps a missing response separate from an incorrect mapping. There is no leader to compare with the contract. A transport failure, stale cache or incomplete log could explain the gap, and this window does not infer which occurred. The caller must attach the exact response and the route from the same run before the local audit packet can be useful.

Using the result in your application

Download the exact sample requests and saved observations below.

Read the example notes

Expected labels are original fixture checks and stay outside the request sent to Jev. The observation file preserves the complete probabilities and the hash of each exact request. Editing a task, rubric or candidate requires a fresh response; a matching sample name does not make an older answer current.

Using the result in your application

The JavaScript runner makes one independent SDK request for a chosen sample ID.

Read the example notes

It reads TYPESAFE_API_KEY privately from the environment and has no automatic retry. The Python consumer operates offline: it checks the rubric, labels, unique sample identities, request hashes and valid probability distributions before selecting a fixed local preview. A tied leading result remains pending for review.

Source and example scope

The page reports measured adapter round trips and estimated input cost for the recorded calls.

Read the example notes

Those values describe these requests, not a provider latency guarantee or a billing statement. Six deliberately authored fixtures can check the stated behavior, but they do not establish accuracy on an unseen dataset. Keep disagreements and missing-input examples when testing an integration.

Runnable offline example

Replay one request and inspect the saved preview.

View the code example
Replay commands
npm install @typesafe-ai/sdk@0.6.0
# Set TYPESAFE_API_KEY privately in your shell.
node run-choice-sample.mjs samples.json billing
python preview_choice_workflow_v2.py samples.json saved-api-results.json
Scope of this demo

Where does this decision fit?

Useful with explicit supplied evidence

  • Audit a supplied decision-to-route contract against a synthetic model record and the reported application route. Distinguish faithful mapping, an application override and missing contract or result. Return a pending audit packet; do not classify the underlying ticket, authenticate logs or route real work.
  • Inspecting the original task and a fixed output before integrating external tools.
  • Comparing an explicit expected label with the complete recorded distribution.

Needs separate implementation or evidence

  • Inferring facts absent from the supplied input.
  • Treating a semantic selection as authorization to act.
  • Executing an external action or generalizing six fixtures into an accuracy claim.
Example FAQ

Questions about this original decision workflow.

Keep relevance, compliance and execution separate.

Is this another ticket classifier?

No. It audits an already supplied model record and the application route against an explicit mapping contract.

Which responses are actual API observations?

The six outer Jev audit calls and the edited browser request are real. The inner probability records are explicitly synthetic examples.

Does Faithful mapping prove the model label is correct?

No. It says only that the reported application route follows the supplied contract and unique inner leader.

Can the audit authenticate application logs?

No. These are caller-supplied records. Actual log integrity and matching run identities need separate verification.

Does this reproduce the source financial analysis?

No. It uses original support-queue handoffs and makes no source performance, company assessment or financial recommendation.

Try a task from your own workflow

Change the supplied task or evidence here, then inspect the live response. Playground supports another explicit decision.