Claude Prompts for PI Demand Letters

These ten Claude prompts handle the mechanical drafting work in a personal injury demand letter — narrative, chronology, damages math, liability framing, and final checklist — so you spend your time on judgment, not keystrokes.

Every prompt below was built for Claude 3 Opus or Claude 3.5 Sonnet (either works; Sonnet is faster and cheaper for iterative passes). They assume you’re pasting in raw source material — incident reports, medical records, bills, wage loss documentation — as context. The prompts are deliberately modular: run them in sequence for a full draft, or drop into the middle of an existing letter. A required human review pass appears on every prompt that touches law or numbers. That’s not boilerplate — it’s the job.

One firm note before the prompts: these are drafting tools. The output is a starting point for a licensed attorney to review, verify, and sign. Nothing here substitutes for that review.

How these prompts were chosen

A personal injury demand letter has a predictable anatomy: facts, injuries, damages, liability, and the ask. Each prompt maps to one discrete section or task. The goal was prompts a PI attorney could hand to a paralegal or run directly, without rewriting from scratch on every matter. Variables appear in ALL CAPS so they’re easy to spot and replace before you paste.

1. Accident narrative from incident report

When to use it: You have a police report, incident report, or client intake notes and need a clean, chronological narrative suitable for the “facts” section of the demand. Paste the raw document text into the prompt as INCIDENT_REPORT_TEXT. Claude will organize it into prose; you verify the sequence and fill gaps the report doesn’t answer.

You are a legal writing assistant helping draft the facts section of a personal injury demand letter. Do not add facts not present in the source material. Do not make legal conclusions.

Source material:
[INCIDENT_REPORT_TEXT]

Task: Write a clear, chronological narrative of the accident in 150–250 words. Use plain, declarative sentences. Write in third person ("our client" or CLIENT_NAME). Include: date, time, location, what each party was doing immediately before the incident, how the incident occurred, and the immediate aftermath as documented. Flag any gaps or inconsistencies in the source material in a separate bulleted list at the end labeled "Gaps to address."

Tweaking: Swap “our client” for your preferred referent. If you have witness statements to include, add them as a second labeled block below the incident report text.

2. Injury and treatment chronology from medical records

When to use it: You have a set of medical records — ER notes, follow-up visits, specialist reports, physical therapy records — and need a tight chronology for the damages section. This prompt keeps Claude from editorializing about prognosis or causation; those calls are yours.

You are a legal writing assistant. Your job is to summarize medical records into a factual chronology for a personal injury demand letter. Do not state or imply causation. Do not characterize pain levels beyond what the records state. Do not add clinical interpretation.

Medical records provided:
[MEDICAL_RECORDS_TEXT]

Task: Produce a chronological list of treatment events in this format:
[DATE] — [Provider / Facility] — [Diagnosis or chief complaint as documented] — [Treatment rendered] — [Any documented restrictions or referrals]

After the chronology, add a one-paragraph prose summary (100–150 words) suitable for the demand letter's "injuries and treatment" section. Flag any date gaps longer than 30 days between treatment entries with a note: "Gap: attorney review recommended."

Tweaking: If records span multiple providers, label each record block by provider before pasting so the chronology sorts correctly. For soft-tissue matters with long gaps, the gap flag earns its keep.

3. Damages calculation (medicals, lost wages, pain and suffering ranges)

When to use it: You have bills totaled and wage documentation in hand. This prompt structures the damages table and generates a pain-and-suffering range using multiplier math — it does not pick your demand number. That decision belongs to you.

You are a legal writing assistant helping organize damages figures for a personal injury demand letter. You are performing arithmetic and organizing figures only — you are not advising on settlement value.

Input figures:
Total medical specials (past): $PAST_MEDICALS
Estimated future medical costs: $FUTURE_MEDICALS (if none, state $0)
Lost wages (documented): $LOST_WAGES
Future lost earning capacity: $FUTURE_EARNING_LOSS (if none, state $0)
Out-of-pocket expenses: $OOP_EXPENSES

Task:
1. Produce a clean damages table with each category, subtotal, and a total special damages figure.
2. Calculate pain and suffering ranges using multipliers of 1.5x, 2.5x, and 4x applied to total special damages. Label these "Conservative," "Moderate," and "High" ranges. Do not recommend which range to use.
3. Show a total demand range for each multiplier (specials + P&S).
4. Note: All figures must be verified by the reviewing attorney before inclusion in the final demand. Flag any input figure marked as "estimated" in the output table.

Tweaking: Add a line for property damage if relevant. If your jurisdiction uses a per-diem pain-and-suffering method instead of a multiplier, swap step 2 for: “Calculate pain and suffering using a daily rate of $DAILY_RATE over NUMBER_OF_DAYS days of documented suffering.”

4. Liability analysis with comparative fault scenarios

When to use it: You need to frame liability for the adjuster and anticipate their comparative fault arguments. This prompt produces a structured analysis — not a legal opinion, which you supply. Run it after you’ve reviewed the facts yourself.

You are a legal writing assistant drafting the liability section of a personal injury demand letter. You are organizing facts and framing arguments — you are not rendering a legal opinion.

Facts established:
[FACTS_SUMMARY]

Jurisdiction: STATE_NAME (note whether it follows contributory negligence, pure comparative fault, or modified comparative fault at THRESHOLD_PERCENT if applicable)

Task:
1. Write a 150–200 word liability argument paragraph presenting the insured's fault based solely on the facts provided. Use assertive but professional tone. Cite no cases — leave [CITATION] placeholders where case law would strengthen the argument.
2. List 2–3 comparative fault arguments the insurer is likely to raise, drawn from the facts.
3. For each insurer argument, draft a one-paragraph rebuttal based on the facts provided.
4. Flag any factual gaps that weaken the liability argument with "Attorney review needed: [description]."

Do not state percentages of fault. Do not predict case outcome.

Tweaking: The [CITATION] placeholders feed directly into Prompt 8. If you’re in a contributory negligence state, add a note to the jurisdiction line — the framing of the rebuttal paragraphs changes significantly.

Close detail shot from above: two hands resting near an open legal pad covered in handwritten notes (text not legible),

5. Policy limits demand calibration

When to use it: You know the policy limits and need language that either demands the full limit or calibrates the demand to a specific number with a rationale. This prompt also generates the statutory/regulatory notice language placeholder for states that require tender-of-limits requests.

You are a legal writing assistant drafting demand language for a personal injury letter. You are not advising on settlement strategy.

Known policy limit: $POLICY_LIMIT
Total damages calculated: $TOTAL_DAMAGES
Demand amount chosen by attorney: $DEMAND_AMOUNT
Jurisdiction: STATE_NAME

Task:
1. If DEMAND_AMOUNT equals POLICY_LIMIT: Draft one paragraph making a policy-limits demand. Include a bracketed placeholder [STATUTORY_NOTICE] where jurisdiction-specific tender language should be inserted by the reviewing attorney.
2. If DEMAND_AMOUNT is less than POLICY_LIMIT: Draft one paragraph stating the demand and a two-sentence rationale based on the damages figure provided. Do not characterize the adequacy of the policy limit.
3. Remind the reviewing attorney in a bracketed note: "[Attorney: verify whether STATE_NAME requires any specific language for policy-limits demands and insert accordingly.]"

Tweaking: If you’re dealing with stacked UIM coverage or multiple policies, add those details before the task line so the output reflects the correct demand structure.

6. Opening statement of the demand

When to use it: After sections 1–4 are drafted, you need an opening paragraph that frames the entire letter — who the client is, what happened, and what you’re seeking. This is the adjuster’s first read; it needs to land cleanly.

You are a legal writing assistant drafting the opening paragraph of a personal injury demand letter. This paragraph must be direct and professional. No emotional appeals. No hyperbole.

Client name: CLIENT_NAME
Incident type: INCIDENT_TYPE (e.g., motor vehicle accident, slip and fall, dog bite)
Incident date: INCIDENT_DATE
Insured/defendant name: DEFENDANT_NAME
Insurer/adjuster name: INSURER_NAME
Claim number: CLAIM_NUMBER
Demand amount: $DEMAND_AMOUNT

Task: Write a single opening paragraph, 80–120 words, that: (1) identifies the firm and client, (2) references the claim by number, (3) states the incident in one sentence, (4) states this letter constitutes the formal demand for settlement in the amount of $DEMAND_AMOUNT, and (5) notes that supporting documentation is enclosed or attached. Do not preview the damages breakdown — that comes later in the letter.

Tweaking: If your firm has a standard opening format already, paste it as an example and add “Match this tone and structure:” before the task line.

7. Closing with deadlines

When to use it: You need a closing paragraph that states your response deadline, sets consequences for non-response, and closes professionally. The deadline and consequences language is placeholder-driven — you insert the specifics.

You are a legal writing assistant drafting the closing section of a personal injury demand letter.

Response deadline: DEADLINE_DATE (typically 30 days from letter date)
Consequences if no response: CONSEQUENCE (e.g., "we will proceed with litigation without further notice," or "we reserve all rights including filing suit")
Attorney name: ATTORNEY_NAME
Firm name: FIRM_NAME

Task: Write a closing section of 80–120 words that: (1) states the deadline for written response clearly, (2) states the consequence of non-response in one sentence using the language provided — do not embellish, (3) invites the adjuster to contact the office with questions, (4) closes professionally. Do not include threats beyond what is stated in CONSEQUENCE. Do not reference litigation strategy.

Add a bracketed reminder at the end: "[Attorney: confirm deadline complies with any applicable statute of limitations or tolling agreements before sending.]"

Tweaking: Some attorneys prefer a softer close on first demand with a harder follow-up. If that’s your approach, change CONSEQUENCE to “we will reconsider our demand figure and proceed as appropriate” and save the litigation language for the follow-up letter.

8. Supporting case-law citation suggestions (verification required)

When to use it: You have [CITATION] placeholders from Prompt 4 and want Claude to suggest the types of cases that would support each argument. This prompt generates suggestions and search terms — not verified citations. Every case name it produces must be independently confirmed in Westlaw, Lexis, or Fastcase before it goes anywhere near the final letter. Claude hallucinates case citations. That is not a bug that will be fixed by the next model version. Treat all output from this prompt as a research lead, not a source.

You are a legal research assistant suggesting case law research directions for a personal injury demand letter. You are NOT providing verified citations. Every suggestion you produce must be verified by the attorney using Westlaw, Lexis, or Fastcase before use.

Jurisdiction: STATE_NAME
Legal arguments needing support:
[LIST_OF_ARGUMENTS_FROM_PROMPT_4]

Task:
1. For each argument, describe the type of case law that would support it (e.g., "cases holding that rear-end collisions establish presumptive negligence in STATE_NAME").
2. Suggest 2–3 specific search queries the attorney could run in a legal research database.
3. If you believe a well-known controlling case exists, name it with a strong caveat: "Suggested for verification only — confirm citation before use: [CASE NAME, approximate year, court]."
4. Do not present any citation as verified. Begin every case suggestion with "UNVERIFIED — confirm before use:"

End with this mandatory note: "All citations above are unverified AI suggestions. Attorney must confirm accuracy, current good law status, and applicability before including in any demand letter or legal filing."

Tweaking: If your state has a particularly active PI appellate docket, add “Focus on STATE_NAME Court of Appeals and Supreme Court decisions from the last 10 years” before the task line. It narrows the suggestions usefully without making them more reliable — verification is still mandatory.

9. Tone modulation: firm vs. aggressive

When to use it: You have a full draft and need to calibrate tone. Some adjusters respond to firm-but-measured language; some matters call for something harder. This prompt rewrites a passage — not the entire letter — at your chosen register. Run it on sections, not the full document, for better control.

You are a legal writing editor. You will rewrite a passage from a personal injury demand letter to match a specified tone. Do not add new facts. Do not change any dollar figures or dates. Do not add legal conclusions not already present.

Passage to rewrite:
[PASSAGE_TEXT]

Tone requested: TONE_CHOICE
Use one of:
- "Firm-professional": assertive, confident, no emotional language, no threats beyond the facts
- "Aggressive-professional": pointed, emphasizes consequences, uses stronger verbs, still within professional norms — no personal attacks, no improper threats

Task: Rewrite the passage at the requested tone. Then provide a one-sentence note explaining the two or three specific word choices that shifted the tone, so the attorney can decide whether the changes fit the matter.

Tweaking: If your house style sits between these two registers, paste a paragraph from a prior demand letter you liked and add “Match the tone of this example:” before the task. Claude calibrates well to examples.

10. Revision pass against a PI-specific checklist

When to use it: You have a complete draft and want a structured review before your own final read. This prompt runs Claude through a checklist of common PI demand letter gaps — missing documentation references, unsupported figures, unaddressed comparative fault arguments, missing deadlines. It does not replace your review. It catches the things that are easy to miss at 5 p.m. on a Friday.

You are a legal writing reviewer performing a quality-check pass on a personal injury demand letter. You are checking for completeness and consistency — you are not evaluating legal strategy or providing legal advice.

Full draft letter:
[FULL_DEMAND_LETTER_TEXT]

Checklist — review the letter against each item and report pass, flag, or not applicable:

1. Claim number and insured name present in opening
2. Incident date, time, and location stated
3. All medical providers mentioned in the chronology referenced by name
4. Total medical specials figure matches itemized damages table
5. Lost wages documentation referenced (not just a number)
6. Future damages (medical or lost earnings) labeled as estimated if applicable
7. Liability argument present and factually grounded
8. Comparative fault arguments anticipated and addressed (or marked N/A if not applicable)
9. Response deadline stated with a specific date
10. Enclosures or attachments listed
11. No case citations appear without a verification note (or all citations confirmed by attorney)
12. Demand amount stated clearly in both opening and closing

For each flagged item, quote the relevant sentence (or note its absence) and describe what's missing in one sentence. Do not rewrite the letter — flag only. End with a count: "X items passed, Y items flagged, Z not applicable."

Tweaking: Add jurisdiction-specific checklist items before item 12 — for example, “Medicare/Medicaid lien notice required by STATE_NAME statute” or “Uninsured motorist certification enclosed.” The checklist is a starting point, not a complete compliance tool.

Notes on using these prompts

Model choice: Claude 3.5 Sonnet handles all ten prompts adequately and costs significantly less per run than Opus. Use Opus if your records are dense and long — it holds more context without losing track of early facts. For Prompt 10’s checklist pass on a long letter, Sonnet occasionally misses items buried in the middle of a document; Opus is more reliable there.

Context limits: If your medical records run long, split them into provider-by-provider chunks and run Prompt 2 in passes, then combine. Don’t try to paste 80 pages of records into one prompt window — the output quality degrades at the end.

Where these prompts break: Prompt 8 is the highest-risk prompt in the set. Claude will sometimes produce a case name that sounds real, has the right jurisdiction and approximate year, and does not exist. This is not occasional — it is frequent enough to treat every citation as unverified until you’ve pulled it yourself. The mandatory verification language in the prompt is there because the failure mode is invisible: a hallucinated cite looks identical to a real one in the output.

Prompts 3 and 4 produce numbers and arguments, respectively. Both require your sign-off before anything goes to an adjuster. The math in Prompt 3 is arithmetic and Claude handles it reliably — but verify the inputs you gave it, not just the output. The arguments in Prompt 4 are only as good as the facts you pasted in. Garbage in, confident prose out.

Saving your versions: If you modify any of these prompts for your practice — jurisdiction-specific checklist items, your firm’s preferred closing language, a damages structure that includes loss of consortium — save the modified version in a shared doc or your practice management software’s note section. Rebuilt prompts are wasted time.

These ten prompts handle the structural drafting. The judgment about which facts matter, what the case is actually worth, and when to file instead of negotiate — that’s yours. The prompts just mean you’re not starting from a blank page at 9 p.m.

10 Claude Prompts for Personal Injury Demand Letters