Customers now ask you to list every AI tool that touches their data, who approved it, and how it's governed. This free kit helps you answer with evidence, not promises, in an afternoon instead of a week.
Illustrative rows. Wording varies by customer.
We also emailed these links to you. Start with the inventory: most other answers come from it.
A workbook you fill in once and reuse for every questionnaire, plus a short guide for the review that's open right now.
One row per AI tool or AI feature: the owner, what data goes in, whether the vendor trains on it, the retention clause and what it leaves out, subprocessors, what agents can do, and who approved it.
| AI tool or feature | Customer data goes in? | Vendor trains on our data? | Retention commitment | Approved by |
|---|---|---|---|---|
| EXAMPLE: AI support chat | Y | Unknown | [Vendor says zero data retention] | [CTO] |
The AI questions buyers ask most, across governance, training on data, retention, subprocessors, agents, logging, incidents, model sources and human review. Each comes with a short answer, a long answer and the evidence to attach.
| # | Category | Question | Status |
|---|---|---|---|
| 6 | Data use and training | Is our data used to train or fine-tune any AI models? | Not started |
| 11 | Retention | Do your AI vendors commit to zero data retention? Which features are excluded? | Not started |
Twenty documents that back the answers, each linked to the questions it supports and where it usually lives, so you can attach proof when a reviewer asks.
| Evidence | Supports # | Have it? |
|---|---|---|
| Vendor contract clauses: training, retention, notification | 6, 10, 11, 21 | |
| Agent permission matrix | 15, 16, 17 |
Two pages on getting through an open questionnaire: triage in the first hour, how to answer a gap honestly, and the mistakes that cost trust.
Every answer has blanks for your facts, and the instructions say to delete any sentence that isn't true for you.
Is our data used to train or fine-tune any AI models?
Why buyers ask: this is usually the first AI question, and the one where a vague answer stalls a deal.
No. Customer data is not used to train or fine-tune models, by us or by our AI vendors. Our vendor contracts [name clause] prohibit it.
Contract clauses and vendor terms of service excerpts. If a vendor's terms are unclear, the guide says to confirm in writing rather than guess.
At a growing SaaS company, the AI row often goes to the person who knows what engineering actually wired up.
You know which AI tools and features exist. The kit turns that knowledge into answers a reviewer accepts.
Your first AI inventory, with the columns buyers ask about already laid out.
A deal waiting on a security review needs AI answers that hold up on the follow-up call.
Yes. You get the full workbook and the guide. We ask for your email so we can send the links and a few short follow-ups.
Usually not for AI questions. A SOC 2 report is an attestation about your security controls in general. Reviewers asking about AI want to know which tools touch their data, how vendors use it, and who approved each tool.
No. The Cloud Security Alliance publishes a free AI security questionnaire, the AI-CAIQ, and the guide points you to it. This kit uses its own wording and is not affiliated with CSA.
No kit can promise that. It helps you document what you do and answer accurately. It is a starting-point template, not legal, tax, insurance, or audit advice.
AI governance software from High Intent Digital. It keeps your AI inventory, policies, acknowledgments and evidence current in one place, so the next questionnaire starts from answers you already have. Bring your vCISO, or we'll introduce one.
Fill in the kit this week, and the next questionnaire takes an hour.
Get the free kit