The AI rollout

Policy before tools

Staff were already curious about AI, and some were already using free tools on their own. For an agency that handles sensitive client information, that's a real risk. So the first thing we built wasn't a training. It was a set of rules.

I helped start the IMPACT AI Advisory Committee, co-facilitated with our IT director, so staff and leadership had one place to set direction. I wrote the AI acceptable use policy and a client confidentiality policy before any licenses went out.

Picking a platform

I evaluated more than seven AI platforms against data privacy standards that made sense for a community action agency. We landed on Claude and rolled out licenses in stages. We're now past 100.

Teaching people to use it well

A license doesn't do much on its own. I built an ongoing training series that covers how to write a good prompt, how the tools compare and how to use them in real work, with facilitator guides so other people can teach it too. The first session drew 30 RSVPs in a week, more than any training we'd run before.

The sessions now run as part of our regular rotation: Tuesday Tech Talks, Feature Focus Fridays and Tech Treks.

What I'd tell another nonprofit

Write the policy first, even if it's short. People are more comfortable trying something new when they know where the lines are. And plan for the training to take longer than the rollout. The licenses are the easy part.