Verified against ChatGPT · 2026-08-11
Summarize what you already know about a legal question so a lawyer's research starts from an organized baseline
Organizes your team's existing knowledge, prior counsel notes, and open questions about a legal issue into a structured research brief that flags exactly what's assumed versus confirmed, so outside counsel's actual research time isn't spent re-deriving what you already had.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You're organizing what my team already knows, suspects, or has been told about a legal question into a research brief — this is a baseline document to hand to a lawyer doing the actual research, not a substitute for that research. WHAT WE ALREADY KNOW OR HAVE BEEN TOLD Prior outside counsel said in 2023 our standard NDA's non-solicit clause was enforceable in our home state; team has since assumed this applies to a new state we're expanding into. THE QUESTION WE NEED ANSWERED Can we enforce the existing NDA's non-solicit clause against a departing employee now based in the new state? SOURCE OF EACH PRIOR NOTE (prior counsel, internal team, general reading) 2023 advice came from outside counsel via email; the 'applies to new state too' assumption came from an internal Slack thread, no lawyer involved. DEADLINE OR DECISION THIS RESEARCH FEEDS INTO Need to decide by end of month whether to send a cease-and-desist letter to the departing employee's new employer. HOW TO ORGANIZE THIS First, restate the core question precisely — often what's actually blocking a decision is narrower than the general topic it's filed under, and getting this restatement wrong wastes a researcher's first hour. Second, go through everything I've told you and sort it into three buckets: confirmed by prior qualified counsel (cite which advice and when, if you know it), assumed or inherited from general reading or team folklore (things believed but never actually confirmed by a lawyer), and directly contradictory information (if two things I told you conflict, surface the conflict rather than picking the one that sounds more authoritative). Third, list what's specifically still unknown and would need actual legal research to resolve — be precise about the gap, not just "more research needed" in general. WHAT NOT TO DO Never attempt to answer the core question yourself, even provisionally or "just to get started" — that is precisely the work this brief is meant to hand off, not preempt, and a plausible-sounding provisional answer in this document risks being mistaken for a real one by someone skimming it later. Never upgrade something in the "assumed or inherited" bucket to "confirmed" based on how often it's repeated in what I gave you — repetition inside a company is not verification. OUTPUT FORMAT 1. Restated core question. 2. Confirmed-by-prior-counsel table: item / source / date if known. 3. Assumed-or-inherited list: item / where it seems to have come from / why it's not confirmed. 4. Contradictions found, stated plainly. 5. Specific research gaps, phrased as precise open questions. 6. A closing line stating this is a research-organization draft only, not legal advice or a legal conclusion, and that the actual question must be researched and answered by a qualified lawyer.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The three-bucket sort — confirmed by prior counsel, assumed or inherited, and contradictory — targets a specific and expensive failure mode in how legal knowledge actually degrades inside a company: a piece of advice given by a lawyer in one specific context (a particular state, a particular year, a particular fact pattern) gets repeated informally until the team treats it as a general, portable rule, and by the time it reaches a new context nobody remembers it was ever conditional. GPT-5.1 is well suited to surfacing this pattern because it can read a mixed pile of notes and messages and separate what was actually attributed to counsel from what was inferred or assumed along the way, but only if explicitly instructed not to smooth the two together into one confident-sounding narrative, which is what a plain summarization request would naturally do. The instruction never to upgrade a repeated assumption to 'confirmed' status directly targets the mechanism by which company folklore acquires false authority — frequency of repetition inside internal messages has zero bearing on whether a lawyer actually verified the claim, and a model that let repetition count as evidence would launder an unverified assumption into something that reads as settled. The hard rule against attempting even a provisional answer to the core question exists because the entire value of this document is in making the handoff to a real lawyer cleaner and faster, not in trying to shortcut it — a provisional AI answer sitting at the top of a research brief is the single most likely thing to get copy-pasted into a decision memo under time pressure, which is exactly the scenario this prompt is designed to prevent.
What you get back
RESTATED QUESTION: Is the existing NDA's non-solicit clause, originally advised as enforceable in our home state, also enforceable against a former employee now in the new expansion state? CONFIRMED BY PRIOR COUNSEL: Non-solicit enforceability in home state — source: outside counsel email, 2023. ASSUMED OR INHERITED: That this advice extends to the new state — source: internal Slack thread, no counsel involved, never actually confirmed. RESEARCH GAP: Whether the new state's law treats non-solicit clauses the same way as the home state. This is a research-organization draft only — a qualified lawyer must research and answer the actual question.
Verified against
ChatGPT GPT-5.1 · 2026-08-11
Changelog
- 2026-08-11 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
