A six-person workforce program ran AI-assisted intake for 90 days. What worked, what didn’t, the actual numbers, and the three things they’d warn another team about before starting.
Disclosure isn’t a yes-or-no question. It depends on what AI did, how much human review there was, and what the participant has a reasonable expectation of. A three-tier framework you can adopt this week.
Federal monitors won’t disallow AI costs because you used AI — they’ll disallow them because you can’t show how the cost benefited the funded activity. A working documentation template you can paste into your file.
Case notes eat the day. AI can take 20–40% of that time back — if you build the workflow carefully. A four-step pattern, the prompts that hold up, and the guardrails non-negotiable for workforce settings.
You don’t need a strategy deck to start with AI. You need a 30-day pilot with one workflow, one tool, two people, and an honest scorecard. Here’s the plan.
Most AI case studies you read are sanitized. The pilot worked. The participants loved it. The metrics moved. We’ll publish field reports differently — here’s the format and why it matters.
Most AI ethics frameworks are written for tech companies. Here’s an honest, actionable framework for small nonprofits that have to live with the consequences — four real ethical questions and how to think about them.
If you operate under federal awards, AI tool costs raise specific Uniform Guidance questions — allowability, allocability, classification, reporting, documentation. Five questions to answer before you charge anything.
Workforce programs face specific AI questions that nonprofit advice doesn’t answer. Here’s where AI actually helps case managers, job developers, and program leads — and where it absolutely doesn’t.
Where to actually start with AI when you run a small nonprofit — not the hype, not the warnings, but the four decisions that determine whether AI helps your mission or distracts from it.