This page lists three turns, the messages you send to the AI one after another in the same chat, and then a check of the answer. Keep it open next to your chat. Each turn shows the next thing to paste, with the values you entered already filled in. Nothing is sent from this page, and your conversation stays in the AI tool you are using.
Paste the memo as it would go out, with the header, distribution list, and any privilege legend intact.
The named person who reads this, their role, how much time they will give it, and how they take legal advice.
The specific decision this memo has to enable, and the date it has to be made.
Different later readers mine for different sentences. The regulator wants dated knowledge; plaintiff's counsel wants admissions.
Who it is from and to, how it is labeled, who will forward it, and whether it mixes legal with business advice.
The law or regulator that would govern if this is ever reviewed; it changes which sentences are dangerous.
Watch for these throughout
The model cannot tell you whether a document is privileged. Dual-purpose memos, non-legal recipients, and investigation summaries are litigated constantly and the tests differ by court. Treat every privilege observation here as a question, not an answer.
Running a memo through this analysis creates a second document about the first one. Decide where the output lives before you generate it.
Advice memos are routinely produced in litigation and enforcement. A sentence written to be candid with the business reads very differently next to what the company said publicly that quarter.
The model pushes toward a clean answer. Some questions honestly do not have one, and forcing certainty into a memo is its own risk. Check that any answer sentence it drafts is one you would sign.
Do not paste memos containing employee or customer personal data, or a live internal investigation, into tools your organization has not approved for that material.