← AI Prompt ToolkitClient work
Interview Quote Extraction
Extract the strongest, most useful quotes from interview transcripts with attribution and timestamps.
When to use it
Use after recording a witness statement, client interview, expert consultation, or any interview where specific quotes will be cited later. Output is organized by quote strength (lead / supporting / color) with verification notes.
The prompt
Copy the prompt below, paste into ChatGPT, Claude, or your firm's preferred LLM, then paste your transcript where the placeholder indicates.
How to use it well
- Provide the Interview Context placeholders before pasting — interviewee name, topic, and use case dramatically improve the quality of extracted quotes.
- Always verify extracted quotes against the original audio (or at least the transcript) before using in court filings or motions — AI can introduce subtle wording errors.
- The "Color quotes" section is useful for assessing whether a witness will present well at trial — captures tone and demeanor in a way fact extraction doesn't.
- For privileged interviews, mark the prompt output privileged and store with the matter file. Do not paste into shared or unsecured AI tools without confirming your firm's AI policy.
Expected output
Quotes organized into Lead / Supporting / Color tiers with attribution, timestamps, context, significance, and verification notes for each.
Related prompts
Key Facts Extractor
Pull facts, evidence references, and legal elements from a deposition into a structured catalog.
Client Intake Summary
Convert a recorded intake meeting into a structured intake summary for the case file.
Contradiction Analyzer
Surface contradictions and inconsistencies within or across deposition transcripts.
Frequently asked questions
Does this work for witness statements outside formal depositions?
Yes. Recorded witness statements, fact-finding calls, and any interview-format recording works the same way. The prompt is content-agnostic — what matters is that the recording has identifiable speakers and quotable substantive statements.
Will it flag privileged or potentially sensitive content?
Yes. Each extracted quote includes a verification-notes field that flags potentially privileged content, factual claims requiring verification, and technical terms that may need explanation. Treat all flags as triage prompts, not legal conclusions.
How accurate are the extracted quotes word-for-word?
High accuracy on the LLM's extraction step, but verify every quote against the source transcript before quoting in court filings or motions. AI can introduce subtle wording errors (a single dropped word, a slightly different verb tense) that matter at the level of legal writing.