AI Vendor Due Diligence Studio with Apoorva Sharma

AppyGuru’s AI Vendor Due Diligence Studio helps procurement, third-party risk, and security teams evaluate AI suppliers without drowning in brochureware. The pack pairs a practical field guide with a questionnaire, contract-starter clauses, a supplier scorecard workbook, and a decision briefing deck so you can reach an honest Go, Conditional, or No-go with owners and review dates.

This AppyGuru digital pack is priced at $19 USD and is designed as a working kit you can adapt in days, not a binder that sits unused. Every artifact ships with an educational notice: these materials are learning aids and starting templates. They are not legal advice, certified compliance, security authorization, or a guarantee that your organization will meet any regulatory requirement. Adapt with qualified counsel and your own risk owners before production use.

## What is inside (file by file)

**01_AI_Vendor_Due_Diligence_Field_Guide.pdf (PDF)** — Four-module field guide with quizzes, worked vendor examples, diligence paths, scoring guidance, and a 90-day plan.

The PDF is written in AppyGuru’s plain-language teaching voice: short modules, concrete examples, exercises, quizzes with answer keys, checklists, and a ninety-day action plan. Use it to align a cross-functional working group before anyone opens the editable templates. Facilitator prompts help you keep meetings focused on decisions, owners, and evidence rather than slogan slides.

**02_AI_Vendor_Diligence_Questionnaire.docx (DOCX)** — Editable Standard/Enhanced questionnaire with SAMPLE answers, version history, and RACI.

The Word workpaper includes visible SAMPLE content so teams know what to replace, plus document control/version history and a RACI table. Instruction callouts mark places that need counsel or owner judgment. Keep SAMPLE markers until a named reviewer signs off; blank fields are useful signals, not failures.

**03_AI_Vendor_Contract_Starter_Clauses.docx (DOCX)** — Counsel-ready clause starters (training, residency, change notice, exit) plus negotiation log.

The Word workpaper includes visible SAMPLE content so teams know what to replace, plus document control/version history and a RACI table. Instruction callouts mark places that need counsel or owner judgment. Keep SAMPLE markers until a named reviewer signs off; blank fields are useful signals, not failures.

**04_AI_Supplier_Scorecard.xlsx (XLSX)** — Multi-sheet register, scorecard bands, conditions, evidence log, and dashboard formulas with sample rows.

The workbook is multi-sheet with an Instructions tab, sample rows, dropdowns where helpful, and operational columns for owners, dates, and status. Treat fictional SAMPLE organizations as teaching data only. Start with the systems or vendors that matter most; connect rows to evidence pointers in your own repository.

**05_Vendor_Decision_Briefing_Deck.pptx (PPTX)** — Go/Conditional/No-go leadership deck with speaker notes.

The deck is built for a short working session or leadership conversation. Speaker notes help a facilitator run the room without an architecture lecture. Replace every SAMPLE line before a real meeting and capture decisions in writing before you adjourn.

## How teams typically use the pack

Recommended order: read the PDF together; assign one owner for the Word workpaper, one for the workbook truthfulness, and one for the briefing or exercise record; replace placeholders that could be mistaken for facts; preserve source links and dates in your evidence space; keep the educational notice attached when distributing adapted copies. When a topic sits outside the team’s authority, record an open question with a named escalation path rather than inventing confidence.

Extended diligence notes: AI vendors change data destiny, model behavior, evaluation opacity, and sometimes autonomy through tools. A SOC report can support a security conversation and still leave training defaults, prompt retention, subprocessor regions, and material-change notice unresolved. Choose Light, Standard, or Enhanced paths from your use context—data class, decision impact, geography, and blast radius—and write the rationale down. Score with Strong / Adequate / Weak / Unknown bands; Unknown is a first-class status that forces an evidence request or a compensating control. Negotiation continues diligence: align product settings with written terms on training use, retention, residency, incident notice, and exit assistance. Brief sponsors with bounded asks—Conditional Go with named conditions beats a vague request to “approve AI.” Revisit at renewal, model-family change, new tool permissions, or new data classes. Cross-sell context: pairs well with AppyGuru AI Governance Foundations when you need inventory habits, and with Agent Incident Response Tabletop when a vendor ships agentic write tools.

Buyer workflow note: this pack is intentionally modular. A facilitator can use the PDF for shared vocabulary, then assign one person to maintain the editable workpaper, one person to validate workbook data, and one person to prepare the briefing or exercise record. Replace every placeholder that could be mistaken for a fact, including names, dates, system IDs, status values, contacts, risk ratings, and sample outcomes. Revisit after incidents, audits, product launches, vendor changes, or new user populations. © AppyGuru — practical AI and governance education.

Additional facilitation tips: keep examples fictional until partners agree what may be used in a live workshop; store system or vendor IDs consistently across PDF exercises, Word forms, Excel rows, and deck slides; calendar the next review before you celebrate the first draft; and treat a blank evidence pointer as unfinished work rather than an implied pass. If leadership asks for a single score, respond with bands, residual themes, and the decision required today. If staff ask whether a paste is allowed, answer with the traffic light, the approved tool, and the escalation contact—not with improvisation. Continuous improvement beats one-time documentation: update the pack artifacts when your reality changes, and keep prior versions in your document repository for traceability. AppyGuru builds original educational kits so teams can practice accountable AI operations with clarity and proportion.