Nature of the service
What Vantage is — and is not
Vantage is a software tool: it helps you explore policy wording, spot possible gaps, and organise next steps when you escalate to an insurer or other bodies. It is not regulated by MAS as financial advice, is not a licensed financial adviser or insurance intermediary, and is not a law firm or substitute for legal counsel. Outputs can be incomplete, outdated, or wrong. They are not the authoritative record of your contract—only your policy documents and your insurer’s written position are. Treat every summary, draft, and checklist as a starting point. Verify material facts against your policy PDF, endorsements, and insurer communications before you act.
If anything here conflicts with your policy PDF, an insurer letter, or professional advice you have received, trust those sources first.
Getting Started
Core workflow and prep checklist
Before you start: prepare your latest policy PDF, specialist details (name and MCR if available), and any rejection letter or claims email.
1) Upload policy PDF and wait for clause-level extraction.
2) Review Findings to identify exclusions, limits, waiting periods, and claim conditions.
3) Open a claim workspace with policy prelinked for grounded triage.
4) Attach supporting docs (bills, rejection letters, specialist memos) in the claim workspace.
5) Use mode-specific actions (pre-submission planning, appeal drafting, payout gap dispute).
User Flows
Recommended path after policy analysis
Analyze policy
Action: Upload insurer PDF and wait for completion.
Output: Findings + cited clauses + policy chat availability.
Risk triage
Action: Review findings and run panel check for specialist.
Output: Tier clarity (`panel`, `extended_panel`, or risk flag).
Create claim
Action: Open claim workspace from the policy report.
Output: Policy prelinked intake and mode-aware next actions.
Ongoing updates
Action: Send turn-by-turn updates and attach supporting docs.
Output: State planner progress, evidence checks, and timeline events.
| Stage | User action | Expected output |
|---|---|---|
| Analyze policy | Upload insurer PDF and wait for completion. | Findings + cited clauses + policy chat availability. |
| Risk triage | Review findings and run panel check for specialist. | Tier clarity (`panel`, `extended_panel`, or risk flag). |
| Create claim | Open claim workspace from the policy report. | Policy prelinked intake and mode-aware next actions. |
| Ongoing updates | Send turn-by-turn updates and attach supporting docs. | State planner progress, evidence checks, and timeline events. |
If you are handling a rejection, start directly in My Claims and switch mode to rejected or partial-approved as needed.
Claim Modes Playbook
What each mode should do
Pre-submission
Confirm panel status, pre-authorisation requirements, and likely out-of-pocket exposure.
Awaiting decision
Track insurer response windows and prepare follow-up evidence proactively.
Rejected
Capture rejection wording, challenge missing clause basis, and generate an appeal draft.
Partial approved
Quantify unpaid gap, identify deduction basis, decide accept-versus-dispute path.
Evidence & Citations
How to interpret response quality
- Evidence-grounded: key assertions are backed by retrievable sources and citation metadata.
- Uncertain: missing facts, conflicting evidence, or unsupported assertions were detected.
- Informational: generic guidance without clause-level grounding for that turn.
- Citation contract fields: source_url, source_ref, and page.
- Retrieval spans insurer policies and regulatory references (MAS/FIDReC) when relevant.
Router Modes
Rollout and enforcement
shadow
Observe routing decisions; no full hard enforcement.
enforce_high_risk
Hard enforcement for high-risk intents (`appeal`, `panel_status`).
enforce
Hard enforcement across all supported intents.
| Mode | Behavior |
|---|---|
| shadow | Observe routing decisions; no full hard enforcement. |
| enforce_high_risk | Hard enforcement for high-risk intents (`appeal`, `panel_status`). |
| enforce | Hard enforcement across all supported intents. |
CLAIMS_ROUTER_V1_MODE=shadow|enforce_high_risk|enforce- Router assigns intent per turn (`coverage_advice`, `panel_status`, `appeal`, `general`).
- Tool policy decides when policy retrieval and panel checks run.
- Evidence verifier and contradiction checks block unsupported conclusions.
- Coverage math is calculator-first; chat avoids ungrounded numeric outputs.
Troubleshooting
Common issues and fixes
Issue: assistant keeps asking for rejection wording even after you said no clause was provided.
Action: paste any exact rejection sentence you have and explicitly state "no policy basis clause cited".
Issue: citations point to storage URLs only.
Action: ask for clause or regulatory web citations in addition to policy excerpts.
Issue: panel status is uncertain.
Action: provide doctor full name + MCR and re-run panel check from Panel Checker.
Issue: uploaded supporting docs are ignored.
Action: attach docs in claim workspace chat and send a message in the same turn so extraction is triggered.
Feedback & Support
Issue capture loop
Use /feedback for bug reports, wrong-answer incidents, UX friction, or feature requests.
For fastest triage include:
1) page URL, 2) what you asked, 3) what happened, 4) what you expected, 5) screenshot if possible.
Privacy & Retention
Data handling policy
Review full details at /privacy.
Session mode auto-expires. Signed-in mode persists until user-initiated deletion.
Use account deletion and claim deletion actions to remove retained data when required.