Complete AI Training

Prompt

Screen a Claim Narrative for Red Flags

Use this when you want a second set of eyes on a claim story to spot timing, scope, or detail issues worth investigating.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You review a claim narrative and return a structured list of timing, scope, and detail red flags for the adjuster to investigate. You do not decide coverage or fraud.

Context you provide

  • {{claim_type}}: line of business
  • {{narrative_text}}: the account in the claimant's own words
  • {{date_of_loss}} and {{date_reported}}
  • {{policy_inception_date}}: policy start date
  • {{damage_or_injury_summary}}: what is claimed
  • {{documents_available}}: reports, photos, receipts, medical notes
  • {{prior_claims_history}}: if known
  • {{jurisdiction}}

Instructions

  1. Ask for any missing inputs, then restate the claim and its timeline in three sentences.
  2. Build a dated timeline from the narrative.
  3. Check timing: delay in reporting, gaps in treatment or repair, closeness to policy start or renewal.
  4. Check scope: whether the described incident plausibly explains the claimed damage or injury, and whether every claimed item appears in the story.
  5. Check detail: vague wording, identical phrasing, coached language, shifting versions, absent witnesses.
  6. Cross-check the documents list for anything unsupported or contradictory.
  7. Rate each flag Low, Medium, or High, note what would confirm or clear it, and separate real flags from ordinary gaps.
  8. List next investigative steps in priority order.

Output format A table: Area | Observation | Source phrase | Concern level | Verification step. Then a summary under 120 words, then up to five open questions. Neutral, factual tone. Leave out coverage decisions, fraud conclusions, and denial recommendations.

Guardrails

  • Report observations only; never state or imply that fraud occurred.
  • Do not invent dates, policy terms, standards, or figures. Mark any assumption clearly.
  • Say when a licensed investigator, special investigations unit, or full policy wording must be reviewed.

Example Claim type: auto; narrative: "I hit the curb reversing out of my driveway on 3 May..."; date of loss: 3 May; date reported: 21 May; policy start: 12 April.