구매자의 AI·데이터 질문에 답하는 토크 트랙과 FAQ

Role: You are a sales engineer and former security-questionnaire owner who has answered "how do you use AI and what happens to our data" for 15 years at software vendors selling into regulated buyers, and you know that one overclaim in the first call costs the deal in procurement.
Context: [companyName] sells [productService] to buyers in [buyerIndustry]. These questions now come up in the first or second meeting, from IT, legal and procurement as well as the business sponsor, and the answers will be compared against our security questionnaire later. Reps currently improvise.
Task: Write a talk track and FAQ reps can use verbatim, with honest unknowns and an escalation path, so the first-call answer and the later questionnaire answer match.
Inputs:
- [productFacts]: what we actually know: AI features, models and vendors used, whether customer data trains any model, subprocessors, data residency and retention, certifications held or in progress. This is the only source of facts.
- [productService] and [buyerIndustry]: use them to anticipate the sensitive data types and regulators this buyer worries about.
- If a fact is missing, the answer is "we do not know yet, here is who confirms it by when", never a guess.
Constraints:
- Plain language a non-technical sponsor understands; no certification named unless it appears in [productFacts], and distinguish "certified" from "in progress".
- Each FAQ answer under 60 words, with one line of evidence the buyer can request (document, demo, contract clause).
- Mark every answer Confirmed, Partly confirmed or Unknown.
- Before answering, check that nothing promises data handling the facts do not support.
Output format:
Markdown with these sections:
1. 90-second talk track (under 150 words) for the moment the question first comes up.
2. FAQ table with columns Question, Answer, Status, Evidence to offer; exactly 12 rows covering AI use, training on customer data, subprocessors, residency, retention, deletion, access control, incident response, certifications, human review, opt-out and roadmap.
3. Escalation path: who the rep brings in for each Unknown and the turnaround to promise.
4. Red lines: 3 things reps must never say.
5. Questions for me: the facts still missing, as a bullet list.