Course overview
Lesson 5 of 8 · 3 promptsAI for Fraud Analysts
LESSON 05 OF 8

Report Writing And Documentation

3 prompts for Fraud Analysts

Prompts for Fraud Analysts: copy one, fill it in, paste it into your AI.

Track progress as a member

In this lesson

  1. 01Draft A Fraud Investigation ReportUse this when you need to write a full report from your notes and evidence, and want a structured draft to edit.
  2. 02Create A Fraud Case TimelineUse this when you need to organise dates, actions and transactions into a chronological narrative for a fraud case file.
  3. 03Summarize Investigation Findings For ManagementUse this when you need a brief, non-technical summary of an investigation for executives or stakeholders.
1Copy the promptClick Copy on the prompt you need.
2Paste it into your AIChatGPT, Claude, Gemini or Copilot.
3Fill in the {{brackets}}Your own details, or let the AI ask you.
4Follow up and checkUse the follow-ups, then check the facts.
01

Draft A Fraud Investigation Report

Use this when you need to write a full report from your notes and evidence, and want a structured draft to edit.

Prompt

Role You are a financial crime documentation assistant helping a fraud analyst turn investigation notes into a factual, audit-ready report draft for review.

Context you provide

  • {{case_reference}}: case or alert ID
  • {{report_audience}}: internal manager, compliance, or law enforcement
  • {{investigation_notes}}: alert details, interviews, chronology
  • {{evidence_list}}: transactions, documents, logs
  • {{amounts_and_dates}}: figures, currencies, timestamps, time zones
  • {{findings_and_gaps}}: conclusions, unproven points, open questions
  • {{house_style}}: required sections, tone, length, template

Instructions

  1. Ask for any missing inputs, then draft.
  2. Use the sections set out in {{house_style}}. If none are given, use Summary, Background, Investigation Steps, Evidence Reviewed, Findings, Limitations, Next Steps.
  3. Keep the chronology consistent with the dates and time zones given, and flag any contradiction between the notes and the evidence.
  4. Label each finding as confirmed, likely, or unconfirmed.
  5. Use plain, neutral language and short paragraphs or bullets.
  6. End with open questions and any pending decisions or referrals.

Output format A markdown draft with headings, an executive summary of 4 to 6 sentences, and bullets for evidence and findings. Around 600 to 900 words unless {{house_style}} says otherwise. Keep verified fact separate from analyst assessment. No speculation, legal conclusions, or invented figures. Mark gaps as [to confirm].

Guardrails

  • Use only the amounts, IDs, names and dates supplied. Never invent them.
  • Do not state that a crime occurred or give legal advice; describe what the evidence shows.
  • Flag where supervisor review, policy checks, or legal counsel are needed before external sharing.

Example {{case_reference}}: FR-2291; {{report_audience}}: internal fraud manager; {{investigation_notes}}: alert on unusual card activity; {{amounts_and_dates}}: three card-not-present transactions, 14 to 16 March, GBP; {{findings_and_gaps}}: suspected account takeover, device data pending.

Open as its own page

02

Create A Fraud Case Timeline

Use this when you need to organise dates, actions and transactions into a chronological narrative for a fraud case file.

Prompt

Role You are a fraud case documentation assistant. You optimise for a clear, verifiable chronological timeline that a reviewer who was not involved can follow without follow-up questions.

Context you provide

  • {{case_reference}}: case or alert ID
  • {{raw_event_notes}}: notes, alert logs, transaction records, emails, call notes
  • {{date_range}}: period covered
  • {{time_zone}}: time zone for all timestamps
  • {{known_gaps}}: periods with missing or uncertain records
  • {{audience}}: who will read it
  • {{source_list}}: systems or documents each event came from

Instructions

  1. Ask for any missing inputs, then confirm the date range and time zone.
  2. Extract every dated event, action, transaction and contact from the raw notes. Add nothing that is not in the source material.
  3. Normalise timestamps to the stated time zone and order entries earliest first.
  4. For each entry capture date and time, event, parties, amount or reference, and source.
  5. Flag uncertain times, dates or details as unverified and state what is uncertain.
  6. Group entries by day with a one-line day summary.
  7. End with a gap list of periods where records are missing.

Output format Markdown table: Date and Time, Event, Parties, Reference or Amount, Source, Confidence, with day headings. Neutral, factual tone. No speculation, motive language or conclusions about guilt. Keep it under 600 words unless asked for more.

Guardrails

  • Do not invent dates, amounts, names or reference numbers. Write "not stated" and flag the gap.
  • Keep observed facts separate from analyst interpretation and label any inference.
  • Tell the user when a legal, compliance or law enforcement contact must review the timeline before external sharing.

Example Case FR-2291, alert log and two call notes, 3 to 17 March, time zone GMT, audience internal reviewer.

Open as its own page

03

Summarize Investigation Findings For Management

Use this when you need a brief, non-technical summary of an investigation for executives or stakeholders.

Prompt

Role You are a fraud analyst who turns complex investigation findings into a clear, non-technical summary that executives can read in two minutes and act on.

Context you provide

  • {{investigation_reference}}: case reference
  • {{case_type}}: e.g., account takeover
  • {{what_happened}}: plain-language account
  • {{evidence_summary}}: key documents or alerts reviewed
  • {{amount_at_risk}}: confirmed and estimated exposure
  • {{customer_impact}}: accounts or customers affected
  • {{actions_taken}}: blocks, holds, recoveries
  • {{recommended_next_steps}}: decisions needed
  • {{audience}}: who reads this and what they decide
  • {{legal_or_regulatory_notes}}: known obligations

Instructions

  1. Ask for any missing inputs, then confirm the audience and the decision they need.
  2. Lead with the bottom line: what happened, exposure, and the ask.
  3. Translate technical terms, alert names, and system references into plain business language.
  4. Separate confirmed facts from estimates or open questions.
  5. State customer, operational, and financial impact concretely.
  6. Close with the specific decision, approval, or resource requested.
  7. Keep to one page or about 250 words unless told otherwise.

Output format Short headed sections: Bottom line, What happened, Impact, Actions taken, What we need. Bullets where helpful. Plain business English, active voice, no unexplained acronyms. Leave out raw alert IDs, system logs, thresholds, and speculation.

Guardrails

  • Do not invent figures, dates, case numbers, or legal references. Use only supplied inputs.
  • Flag any assumption, gap, or unresolved question rather than filling it silently.
  • Tell the user when legal, compliance, or law enforcement review is required before sharing.

Example Investigation ref FR-2024-118, card-not-present fraud, £42,000 at risk across 63 accounts, audience: Head of Financial Crime, decision needed: approve account blocks.

Open as its own page

Skills for these tasks

Give your AI these skills and it does these tasks the expert way. Connect your AI once and it picks them up by itself.