← Selected work

Ticket Intelligence · Portfolio prototype

Better ticket handoffs.
Every fact traceable.

Turn support notes into structured evidence, identify what is missing, and recommend a queue using a configurable policy. Keep the support team in control.

Explore the demo ↓

Evidence you can inspect

Source quotations support the extracted facts. The original note stays available for review.

Missing details stay visible

Checks flag gaps in scope, impact, errors, and troubleshooting. Unknown information stays unknown.

Routing with a reason

Editable rules connect documented conditions to queues. Recommendations identify the policy and matching rule.

Invite-only · Live AI analysis

Your note. An inspectable handoff.

Have a demo invitation? Enter your access code and a fictional or anonymized support note. The model extracts quoted evidence; the MVP checks the quotations and applies its completeness and routing rules. This is a real API call, not a prepared example.

Ask Alan for an invitation. The code is not saved by this page.

0/2,000 characters. Do not include names, passwords, account details, customer data, or confidential information.

The application does not save submitted notes or results to a database. Hosting and API providers have their own processing and retention policies. Use synthetic data only if unsure.

Access-code protected. Shared usage allowance applies.

What you will see

Quoted evidence, a completeness checklist, advisory priority, and a queue recommendation with its matching rule.

Unknown information stays unknown. If the model returns unacceptable evidence, no result is displayed.

The code is a shared invitation gate, not a user account. This prototype does not diagnose root cause, connect to a ticketing system, or change a ticket.

Try the public examples instead ↓

Curated example explorer

See how evidence changes the handoff.

These are synthetic tickets with hand-authored extraction fixtures. Results were generated by the MVP’s validation and routing code. Selecting an example displays a prepared result; it does not call a live AI model.

Original synthetic note

vpn not connecting user says it worked yesterday

Quoted evidence

Symptom“vpn not connecting”

View structured ticket
Reported symptoms:
- "vpn not connecting"
Reported history / context:
- "user says it worked yesterday"
Documented troubleshooting:
- Not documented.
Original note (verbatim):
vpn not connecting user says it worked yesterday

Showing: Incomplete note · Priority Unknown · Intake / clarification

Category suggestion
VPN
Derived priority
Unknown
Completeness
2/10

Recommended queue

Intake / clarification

Collect missing or uncertain evidence before choosing a technical queue. Readiness gate missing: Affected-user scope, Business impact

Missing information

  • Device or service identity
  • Exact error or status
  • Troubleshooting already attempted
  • Affected-user scope
  • Business impact

Suggested next step

Confirm whether work is blocked, degraded, or nonblocking. Confirm how many users or locations are affected. Capture the affected device/service name and, for VPN, the client/product name. Capture the exact error or status message. Document any steps already performed and their outcomes before repeating work.

Inspect routing decision

Synthetic policy worleyai-demo · version 1.0 · rule fallback-human-triage

    Advisory only. No ticket system is connected and no ticket is changed. Completeness is a checklist score, not a probability of correctness.

    What this demonstrates

    A practical MVP,
    with clear boundaries.

    The invite-only demo uses live model extraction, evidence validation, and configurable routing. Public examples use synthetic fixtures and do not call a model. Saved human review is available in the local application only. Ticket-system adapters, enterprise authentication, and measured operational results are future work.

    Exact source matches help catch unsupported quotations. They do not prove an interpretation is correct. A technician still reviews the recommendation.

    Discuss Ticket Intelligence ↗