AI System Inventory Worksheet
Working document for cataloging every AI tool and AI-powered feature in use across the advancement operation.
Use this worksheet to catalog every AI system and AI-enabled feature in use across your advancement operation. Include standalone AI tools, AI features embedded in existing platforms (CRM scoring, predictive analytics modules, generative drafting assistants), and any third-party services that apply machine learning to institutional data.
Review and update quarterly, or whenever a new AI system is adopted or an existing system changes scope. The completed inventory feeds into risk classification, vendor assessment, and compliance review.
Risk tiers referenced below are defined in the AI Governance Committee Charter: Tier 1 = minimal risk, Tier 2 = moderate risk, Tier 3 = high risk.
Data sensitivity levels: Public — published or freely available. Internal — not public but not regulated. Sensitive — includes PII or data subject to contractual restrictions. Restricted — regulated data (FERPA, HIPAA) or legally privileged.
Decision impact levels: Informational — provides context only, no direct action follows. Advisory — recommends an action a human then takes. Determinative — drives a decision with minimal human override. Automated — system acts without human intervention.
- Institution
- [Institution name]
- Department
- [Department]
- Completed by
- [Name and title]
- Date
- [Date completed]
- Next review date
- [Date]
- Reviewed by
- [Name, title, and date]
System inventory
Complete one row per AI system. Pre-filled rows are examples — replace or remove them as needed.
| System name | Vendor / source | Function | Business process | Data inputs (sensitivity) | Data outputs | Data residency | Training data scope | Decision impact | Human review | Constituent exposure | Regulatory surface | Risk tier | Owner | Last reviewed |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CRM Prospect Scoring | [CRM Vendor] | Assigns propensity and affinity scores to prospects based on giving history, engagement, and demographic data | Prospect identification and portfolio prioritization | Giving history, event attendance, email engagement, biographical data (Internal) | Prospect scores, rank-ordered prospect lists | Vendor cloud (US) | Aggregated (multi-client) | Advisory | Yes | Yes — scores influence solicitation priority | FERPA (school official exception); State law (profiling) | 2 | [Director, Prospect Research] | [Date] |
| Enterprise AI Assistant | [Vendor] | Drafts donor correspondence, proposals, and stewardship reports from staff prompts | Donor communications and stewardship | Staff prompts, templates, institutional boilerplate (Internal) | Draft text for human review | Vendor cloud (US) | Pre-trained (general-purpose LLM) | Informational | Yes | Indirect — outputs shape constituent communications after human review | None | 1 | [Director, Annual Giving] | [Date] |
| Wealth Screening AI | [Data Provider] | Estimates net worth, real estate holdings, and philanthropic capacity using public and proprietary data sources with ML models | Prospect research and major gift pipeline development | Constituent names, addresses, employer data (Sensitive — results include estimated financial data) | Wealth estimates, capacity ratings, giving indicators | Vendor cloud (US) | Aggregated (multi-client + public records) | Advisory | Yes | Yes — influences solicitation strategy | FERPA; State privacy law (financial data, profiling) | 2 | [Director, Prospect Research] | [Date] |
Notes
Document any AI features identified during the inventory that require further assessment, reclassification, or committee review.