Successful AI Implementation
Change Management, Minus the Jargon
How small nonprofits can bring AI into their work in a way that's human-centered, doable, and good for the whole team - not just the one person who figured out ChatGPT.
What is change management, really?
Somebody on your team is probably already using AI. Maybe it's the person who does your social media. Maybe it's you, hesitantly asking ChatGPT to untangle a grant paragraph at 9 PM. While it’s great that you or someone on your staff is getting benefit from AI, here's why that’s a problem for your organization: when AI lives in one person's browser tab, it helps one person. When that person leaves - and in a small nonprofit, someone always leaves - it walks out the door with them.
Change management is how you avoid that. Strip away the consultant-speak and it's just this:
Notice what that definition is about. Not software. Not licenses. Not features. People. The tool is the easy part - you can set up an AI account in an afternoon. People take longer, because people come with real questions: Will this replace me? Do I have time to learn this? Is it even ethical? Those questions deserve real answers, not a pep talk.
There's a whole profession built around this, with named methods like Prosci's ADKAR model and John Kotter's 8 steps. They're worth a look someday. You don't need them to start. For an organization of ten people, change management comes down to four habits:
- Talk before you act. People support what they helped shape and resist what got dropped on them.
- Start small and prove it. One workflow, one win, one story to tell.
- Give people time and permission. Learning can't be the eleventh thing on a ten-item to-do list.
- Write things down. If it lives in one person's head, it's not a system. It's a risk.
That’s the whole game. The rest of this page shows you how to run it.
Why bother? Because using AI and benefiting from AI are two different things.
The numbers on this are striking. A December 2025 survey of 346 nonprofits found that nearly every organization has started using AI - and almost none of them feel it has changed what they can accomplish.
Read those first two numbers together. 92% adoption. 7% real impact. That gap is not a technology problem - the 92% and the 7% have access to the same tools. The gap is everything this page covers: talking to your team, picking a smart first project, writing things down, and making AI something your organization does rather than something one person does.
And that ad hoc number matters most for small teams. When 81% of organizations keep no record of what works, every departure means starting over. You already know that story. It's the donor report nobody can run because the person who built it left in March.
Who should be involved?
In a ten-person organization, nobody gets a new job title out of this. These are hats, not hires - and yes, some people will wear two. The minimum viable team is the first two roles. The rest make it stick.
The Sponsor
Says out loud, more than once: "This matters, and I'm making time for it." Protects the hour a month it takes. Uses the tools personally, visibly. If the ED treats AI as someone else's project, everyone else will too.
The Champion
Keeps the effort moving day to day: runs the check-ins, updates the tracker, answers "how did you do that?" questions. Pick someone patient and liked over someone technical and busy. Enthusiasm transfers. Expertise can be learned.
The Early Explorers
They try things first, break things safely, and produce the stories that convince everyone else. Ask for volunteers - never assign this. Drafted explorers become quiet resisters.
The Healthy Skeptic
Skeptics ask the questions you need answered: about privacy, quality, ethics, and whether this is just another shiny thing. A skeptic who feels heard becomes your most credible advocate. A skeptic who feels ignored becomes the reason it fails.
The Board
One page, twice a year: what you're piloting, what it costs, what it's saving, what your guidelines say. Framed right, this is a leadership story - "we're building systems that don't depend on any one person" - and boards love that sentence.
Everyone Else
Nobody should learn about the AI effort through rumor. Everyone hears about it at the same staff meeting, everyone can raise concerns, and everyone eventually gets invited in. "Whole organization" is the goal, remember?
Five phases. Check them off as you go.
This is the actual work, in order. Click each phase to open it. The checkboxes save in your browser, so you can come back to this page and pick up where you left off.
Before anyone opens an AI tool, have the conversation. The goal is simple: everyone hears the same "why," and everyone gets to say what worries them. Fear that goes unspoken doesn't disappear - it goes underground.
Resist the urge to fix everything. Pick one task that's frequent, annoying, and low-risk - a monthly board report, meeting minutes, first drafts of thank-you letters. Not your grant application due Friday. The pilot's job is to produce a win and a story.
This is the step everyone skips, and it's the one that turns a private experiment into organizational change. People aren't persuaded by tools. They're persuaded by watching a coworker they trust save two hours.
Now AI moves from "the pilot team's thing" to "how we work." The secret ingredient is a tiny, boring, recurring habit: a 15-minute weekly check-in where people share what they tried. Small and consistent beats big and occasional, every time.
Change that isn't maintained slowly evaporates. The good news: maintenance is light. This is also where your work becomes turnover-proof - the whole point for a small team.
Keeping track of it all (this is where Notion comes in)
Everything above produces notes: ground rules, pilot results, prompts, wins, decisions. If those scatter across inboxes and sticky notes, you've rebuilt the exact problem you're solving. You need one shared home base - a place anyone on staff can open and see the whole AI effort at a glance.
Notion is a good fit for this: it's free to start, easy to organize, and friendly to non-technical folks. But hear me on this - the habit matters more than the tool. If your team lives in Google Workspace, shared Docs and a Sheet work fine. Use whatever your team will actually open.
Here's the starter setup. Six pages, about an hour to build:
- AI Ground Rules. Your one-page policy: approved tools, what never gets pasted in, and "AI drafts, humans review." The most-linked page in the workspace.
- Pilot Tracker. One row per experiment: the task, who's testing it, time before, time after, status, and a link to the documented workflow.
- Prompt Library. Every prompt that works, with a note on what it's for. This is your organization's AI memory - the thing that survives turnover.
- Wins Log. A running list of hours saved and problems solved. Feeds your board brief, your funder conversations, and your own morale.
- Decision Log. One line each time you decide something: "March: chose Tool X over Tool Y because of privacy terms." Future-you will be grateful.
- Questions & Answers. A judgment-free page where anyone can post "how do I...?" and anyone can answer.
What the Pilot Tracker looks like in practice:
| Task | Owner | Before | After | Status |
|---|---|---|---|---|
| Monthly board report draft | Dana | 4 hrs | 90 min | Documented |
| Meeting minutes & action items | Luis | 1 hr/mtg | 15 min | Documented |
| Thank-you letter first drafts | Priya | 2 hrs/wk | testing | In pilot |
| Volunteer shift reminders | - | 45 min/wk | - | Up next |
Two rules keep the home base alive. First, the Champion owns it - one named person, or it becomes everyone's job, which means no one's job. Second, if a workflow isn't in the workspace, it isn't done. That sentence, repeated kindly and often, is what turns knowledge in people's heads into systems your organization owns.
The potholes
These are the most common mistakes that can easily tank your AI efforts. Click to see how to steer around each one.
A two-minute readiness self-check
Answer honestly - nobody's grading this, and the results live only in your browser.
You don't have to do this alone.
I'm MJ. I spent 35 years in technology, lived in a monastery, and served as a jail chaplain - and somewhere in all of that I learned that the hardest part of any change isn't the tool. It's helping people feel safe enough to try. That's the work I do now with small nonprofits: hands-on AI training that starts small, respects your budget, and builds systems your team can run without me.
If this page raised questions, or you'd like a guide for your first pilot - let's talk.
Send MJ a Message
This article was a collaboration between MJ & AI.
