Adopting AI safely and realistically
A practical process for putting AI to work without introducing unnecessary risk.
Good first use cases
Begin with repetitive, well-understood work where errors can be detected and corrected.
- Drafting donor and member acknowledgments
- Producing meeting summaries
- Answering routine questions through a chatbot
- Segmenting outreach lists
- Organizing unstructured feedback
A safe adoption process
- Govern — assign responsibility and define acceptable use.
- Map — document purpose, users, data flows, and failure modes.
- Measure — test accuracy, privacy, security, and user impact.
- Manage — prioritize risks, apply controls, and improve over time.
Keep people in charge
AI can produce drafts, classifications, summaries, and recommendations. Outputs can be incomplete, inaccurate, or inconsistent, so results must be evaluated against the source information and the business goal. AI should not make unreviewed high-stakes decisions about people, finances, or compliance.
This guide is provided for general information and is not legal, tax, or compliance advice. Specific requirements depend on your organization and jurisdiction.
Related guides
Website or custom web application?
How to tell whether your organization needs a website or a purpose-built application.
Building accessibility in from the start
Why accessible sites are better sites — and what we do about it on every project.
Why a website needs a care plan
A website is an operational asset. Here is what keeping it healthy actually involves.
Book a discovery call.
Tell us about your organization. We will get back to you within one business day to schedule a discovery call.
Let’s build the thing your organization actually needs.
Tell us about your organization. We’ll reply within one business day to schedule a discovery call.