Skip to content
Nonprofit Solutions AIMission-first AI field lab

Nonprofit AI field guide

Reviewed July 21, 2026

Govern vendor and model change

Keep a useful workflow reviewable when products, models, terms, and integrations change.

01

Maintain a change register

Record the service, plan, model or feature, account owner, approved use, data boundary, configured controls, evaluation set, known limitations, and source-review date.

Subscribe to official product, security, privacy, and contract notices where available. Marketing announcements are not a change-control process.

02

Define retest triggers

Retest when the model, provider, plan, integration, prompt pattern, data source, user population, consequence, policy, law, or organizational owner materially changes.

Use the saved representative cases and review rubric. Compare quality and failure behavior, not only speed or a new feature demonstration.

03

Keep an exit path

Document how to disable the workflow, remove access, export needed records, delete retained content where supported, notify users, and return to the human-owned fallback.

Avoid critical work that only one provider, contractor, account, prompt library, or undocumented integration can operate. Portability is part of mission continuity.

Monthly field note

Get practical guidance in your inbox.

Monthly field guidance for mission fit, data boundaries, human review, and bounded AI pilots.
Required