30 ChatGPT Prompts for Personal Injury Lawyers That Speed Up Intake

A personal injury lead goes cold fast. The caller has an injury, a wrecked car, and a list of firms. Whoever answers first and sounds competent usually signs the case. Meanwhile your team is buried in medical records and demand letters.

That is the squeeze these ChatGPT prompts for personal injury lawyers are built for. Below are 30 prompts you can paste today, grouped by the tasks that slow a PI practice down: screening, liability review, record summaries, demand letters, client updates, research framing, and referrals.

This article is not legal advice. These prompts produce drafts, not filings. AI models invent citations and misstate law with total confidence. Therefore a licensed attorney must review every output before it reaches a client, an adjuster, or a court.

The intake bottleneck that costs PI firms signed cases

Personal injury is a speed business. Because the claimant is usually calling several firms in one sitting, the delay between the call and a human response decides who signs.

Research on lead response has consistently found that contact rates collapse as response time stretches from minutes into hours. The Harvard Business Review study on lead response time remains the most cited example.

So the bottleneck is rarely legal skill. Instead it is throughput. Three things eat the day:

• Screening calls that were never going to be cases. Someone still has to take them.
• Reading medical records to find the same six facts. Dates, providers, diagnoses, treatment gaps, prior injuries, and costs.
• Drafting documents with a fixed shape. Demand letters, status updates, referral letters.

None of that requires judgment to start. However, all of it requires judgment to finish. That gap is where these ChatGPT prompts for personal injury lawyers belong.

Ground rules: what you can and cannot put into ChatGPT

Before the list, the guardrails. PI files hold medical history, which is among the most sensitive data your firm touches.

Redact every identifier first

Replace real details with placeholders before you paste anything.

• Client name becomes “the claimant”
• Date of birth becomes [DOB]
• Provider and employer names become “the treating hospital”, “the employer”
• Claim, policy, and case numbers become [REDACTED]

The model drafts just as well from structure. In short, it does not need the real names.

Turn off training on your chats

Disable the data control that lets your conversations improve the model. Better still, use a business tier account where that is off by default and your firm controls retention.

Never ask for a case value

Do not ask ChatGPT what a case is worth, whether liability is clear, or what a jury would award. Those are judgment calls tied to venue, facts, and experience. Because a model will answer confidently anyway, the answer is genuinely dangerous.

Verify every citation

Attorneys have been sanctioned for filing AI generated cases that did not exist. Therefore check every authority against the primary source, every time.

30 ChatGPT prompts for personal injury lawyers

Every one of these ChatGPT prompts for personal injury lawyers works the same way. Replace the bracketed placeholders with your redacted facts. In addition, add a tone instruction when you want the draft to sound like your firm.

First call screening (prompts 1 to 5)

1. “You are helping a personal injury intake specialist. Write 12 screening questions for a first call about a motor vehicle collision. Cover: date, mechanism, injuries, treatment so far, police report, insurance, and prior claims. Plain language.”

2. “Turn these raw call notes into a structured intake summary with sections: incident, injuries, treatment, insurance, employment impact, and open questions. Notes: [PASTE REDACTED NOTES]”

3. “From these notes, list the missing facts we would need before evaluating whether to open a file. Output as a follow up checklist. Do not assess the merits. Notes: [PASTE REDACTED NOTES]”

4. “Draft a polite call script for declining a matter that falls outside our practice areas, offering a referral and no legal opinion.”

5. “Rewrite this intake questionnaire at a sixth grade reading level in English and Spanish, keeping every question’s meaning intact: [PASTE QUESTIONNAIRE]”

Liability quick assessment framing (prompts 6 to 10)

6. “List the categories of evidence typically gathered after a rear end collision, organized by source: scene, vehicle, medical, witness, and official records. Educational checklist only.”

7. “From this police report summary, extract a neutral fact timeline with times, positions, and stated observations. Do not draw conclusions. Report: [PASTE REDACTED SUMMARY]”

8. “Draft a list of questions to ask a witness about a slip and fall, focused on conditions, notice, and timing. Neutral and non leading.”

9. “Explain comparative negligence in plain language for a client handout, noting that rules vary by state and this is general information.”

10. “Review these intake notes and flag any internal inconsistency in dates, locations, or described injuries. Do not add facts. Notes: [PASTE REDACTED NOTES]”

Medical record summarizing (prompts 11 to 15)

11. “Summarize these redacted medical records into a chronological treatment timeline: date, provider type, complaint, findings, and treatment. Flag any gap longer than 30 days. Records: [PASTE REDACTED RECORDS]”

12. “From this treatment timeline, build a provider list with visit counts and date ranges. Output as a table.”

13. “Extract every diagnosis code and plain language diagnosis mentioned in these records, deduplicated, with first mention date. Records: [PASTE REDACTED RECORDS]”

14. “Identify references to prior injuries or pre existing conditions in these records, quoting the relevant line for each. Do not interpret significance. Records: [PASTE REDACTED RECORDS]”

15. “Turn this billing summary into a table of provider, service dates, billed amount, and paid amount, with a total row. Billing: [PASTE REDACTED BILLING]”

Demand letter drafting (prompts 16 to 20)

16. “Draft a demand letter skeleton with sections: introduction, liability facts, treatment summary, damages, and closing. Use bracketed placeholders for every fact and figure. Do not invent numbers.”

17. “Rewrite this liability section to be tighter and more chronological, keeping every factual assertion identical: [PASTE SECTION]”

18. “Convert this treatment timeline into a narrative damages paragraph describing the course of care in plain, non dramatic language. Timeline: [PASTE TIMELINE]”

19. “Review this draft demand for unsupported assertions. Flag any claim without a corresponding record citation. Do not add content. Draft: [PASTE DRAFT]”

20. “Draft a cover email transmitting a demand package to an adjuster, professional and brief, listing the enclosed exhibits: [LIST]”

Client update templates (prompts 21 to 24)

21. “Write a warm 120 word email telling a client their case is progressing, no action is needed, and we will update them next month. Do not predict timing or outcome.”

22. “Draft an email explaining that we are waiting on medical records, why that takes time, and what the client can do to help. Reassuring tone.”

23. “Write a short message requesting a missing document with a specific deadline and one sentence on why it matters.”

24. “Create a monthly status update template with fill in fields: current stage, last action, waiting on, next step.”

Research framing (prompts 25 to 27)

25. “List the general categories of damages typically discussed in personal injury matters, with a plain language description of each. Educational only, no valuation.”

26. “Draft a research question list I should answer about statutes of limitation for a [CLAIM TYPE] claim in [STATE]. Output questions only, not answers.”

27. “Summarize this statute text in plain language, preserving every deadline and condition exactly. Note that I must verify against the official source. Text: [PASTE TEXT]”

Referral out letters (prompts 28 to 30)

28. “Draft a letter referring a matter to another firm, professional and neutral, stating no opinion on the merits.”

29. “Write a client facing letter explaining that we cannot take their matter, with warmth and no legal opinion, including a note about time limits and the need to consult another attorney promptly.”

30. “Draft a thank you note to a referring attorney confirming receipt of a referred matter and next steps.”

Turning a prompt into a repeatable intake checklist

One prompt used once saves minutes. However, ChatGPT prompts for personal injury lawyers turned into a standard save hours every week.

Pick prompt 2, the intake summary. Run it on ten past files and compare the outputs. Then adjust the section list until it matches what your attorneys actually want to see.

Once the format is stable, three things follow:

1. Paste the prompt into your case management notes template, so intake staff use the same one every time.
2. Save your tone instruction as a snippet you append to every drafting prompt.
3. Keep one chat per task type, not per client, so no client data accumulates in any conversation.

As a result, the output stops varying by whoever ran it. That consistency is the actual win, more than the time saved on any single file.

Where AI stops and the attorney takes over

Being explicit about the line protects your clients and your license.

ChatGPT should never decide:

• Whether to accept or decline a case
• What a claim is worth
• Whether liability or comparative fault applies
• Any statute, deadline, or case citation used without verification
• Anything sent to an adjuster or court without full attorney review

The rule is simple. If being wrong would hurt the client, a person decides. ChatGPT can draft the letter explaining the decision. It cannot make it.

Answering after hours calls so leads do not go to the next firm

Here is what the prompt list cannot fix. Every prompt above works on information you already have. None of them pick up the phone.

For most PI firms, that is the real leak. Accident calls arrive at 11 p.m., on Sundays, and while your team is in a deposition. Because the caller is working down a list, voicemail usually means the case goes to the next firm.

Dapta closes that gap with AI voice agents that answer every call, run your screening script, and hand your team a structured summary.

What that looks like in a PI practice

• Answers every call, nights, weekends, and during hearings
• Runs your own screening questions, the same ones in prompt 1
• Speaks natural English and Spanish, with regional accents that sound local
• Books the consultation on your calendar during the call
• Writes the summary to your CRM, so it is waiting when you open the file
• Follows your disclaimers, including that no attorney client relationship forms on an intake call

The pairing is direct. Dapta captures the conversation, and then your ChatGPT prompts for personal injury lawyers turn that capture into a timeline, a records request, and a client email.

For the full picture, explore our related guides:

• AI for personal injury law firms, the practice level overview
• AI intake for law firms, how automated intake runs
• ChatGPT prompts for sales teams, the parent prompt guide
• Best AI tools for law firms, a wider tooling roundup
• AI answering service for law firms, for firms losing after hours calls
• ChatGPT prompts for immigration lawyers, the same approach for immigration practice
• More AI prompt libraries, the full collection

Frequently asked questions

Is it ethical to use ChatGPT in a personal injury practice?

Yes, with safeguards. Bar guidance generally permits generative AI where the attorney keeps competence, protects confidentiality, supervises output, and does not delegate legal judgment. Therefore redact identifiers, verify citations, and read your own state bar’s guidance first.

Can ChatGPT estimate what my client’s case is worth?

No, and you should never ask. Case value depends on venue, facts, insurance limits, and negotiation posture. Because a model will still produce a confident number, relying on it risks real harm to the client.

Is it safe to paste medical records into ChatGPT?

Not in raw form. Extract the text you need, redact identifiers, then paste that. In addition, use a business tier account where chats are excluded from training and retention is controlled.

Will the drafts sound robotic to clients?

They will, unless you direct the tone. Add an instruction like “warm, plain language, under 120 words, no legal jargon” to every client facing prompt. For example, paste one of your own past emails and ask the model to match that voice.

Can AI replace my intake staff?

No. These prompts compress drafting so your staff spend more time on the parts that actually sign cases: listening, building trust, and following up. Someone trained still has to check the output.

What is the fastest place to start?

Start with medical record summaries, prompts 11 to 15. Because the input is already text and the output is verifiable line by line, the risk is low and the time saved is immediate. These ChatGPT prompts for personal injury lawyers pay for themselves on the first file.