September 24, 2026
Nonprofits are excited about AI. They should be. But excitement often comes before clarity, and that gap is where money disappears.
Here's what I'm seeing: organizations rush toward AI because vendors promise it will solve problems.
Sometimes it will. Mostly, it just amplifies the mess you already have, costs more than you expected, and requires someone on staff who actually understands what the tool is doing.
Let me walk you through five truths about AI that every nonprofit leader should know before spending a dollar.
1. AI Costs Money, Not Just Time
This is the hardest one to talk about because vendors specifically market AI as the thing that saves money. It doesn't.
Yes, AI can reduce the time a human spends on a task. But it comes with real costs that people often skip over:
- Monthly subscriptions (ChatGPT Pro is $20, Claude API is metered, specialized tools run $100-$500/month)
- Implementation costs (staff time to set up integrations, test outputs, train team members)
- Staff time to supervise and fact-check AI-generated work (this is non-negotiable)
A nonprofit I worked with spent thousands on AI tools, figured out they were using two of them, and abandoned three. They didn't account for subscription costs in their annual budget. They didn't budget for someone to manage and oversee the outputs. They just saw the "productivity" angle and signed up.
The real math: AI is a tool that costs money upfront and requires someone to use it well. If you don't have a clear problem it solves, or someone with the bandwidth to manage it, you've just added to your expenses.
2. AI Amplifies Your Existing Mess
This one stings because it's true.
Bad data goes in. AI makes it worse, faster. Chaotic processes get documented by AI and stay chaotic, just with prettier formatting. Unclear agreements about who owns what or who's responsible for what don't change because you have AI writing emails. They get faster and more confident.
Here's a real example: a nonprofit had messy donor records (duplicates, inconsistent data entry, fields that meant different things to different people). They wanted to use AI to generate donor reports. The AI did what it was told: it summarized the mess and made it look professional. The leadership team got reports that looked clean but were built on bad data. No one caught it until they realized their donor retention metrics made no sense.
If your data is chaotic, your processes are unclear, or your systems don't talk to each other, AI won't fix that. It will just move faster through the chaos.
Before you adopt AI, audit your actual work. Do you own your data? Can you access it consistently? Do you actually know what you're solving? If the answer to any of those is no, start there.
3. You Need a Human in the Room
AI is not a decision-maker. It's a suggestion engine.
Someone on your team needs to understand what the AI outputs are, why they look the way they do, whether they're trustworthy, and when to reject them. That someone needs judgment, context, and permission to push back.
This is especially critical for nonprofits because your work has stakes. An AI-generated email to a donor should never go out without someone reading it. A grant report built on AI analysis needs a human who knows your mission and your metrics to say, "That conclusion doesn't match what I know to be true." An AI-recommended process change needs a person who understands your actual operations to evaluate it.
The person in the room doesn't need to be a technologist. They need to be someone who thinks critically, knows the work, and is willing to question the output. They also need time in their schedule to do this work.
If you don't have that person, don't deploy the AI tool. You're setting yourself up to act on bad recommendations or send out embarrassing work.
4. AI Readiness Is Boring But Necessary
Before you pick an AI tool, you need to know whether you're ready to use it.
This is the part nobody wants to do because it doesn't feel productive. But it's the most important step.
Ask yourself:
- Do we own our data? Can you extract it, or is it locked in a vendor's system? If it's locked, can you get it out if you need to?
- Can we access it consistently? Is your donor database actually up to date? Is your staff time tracking in one place or scattered across five spreadsheets?
- What problem are we actually solving? Be specific. "We want to be more efficient" is not a problem. "Our grant reports take 40 hours a month and we want to cut that to 10" is.
- Do we have someone with time to manage this? Who will oversee the AI outputs, check for errors, and maintain the process?
If you can answer all four questions clearly, you're ready for an AI conversation. If you can't, you have work to do first. That work is boring. Do it anyway.
5. Beware the Vendor Pitch
Vendors sell AI as a solution to problems you don't have.
A vendor will call and say your nonprofit needs AI for donor segmentation. Maybe you do. Maybe you don't. Maybe you need better data entry first. The vendor isn't evaluating your actual situation. They're selling AI because that's what they sell.
Your job is to own the decision. Not the vendor's job. Yours.
Before you talk to a vendor, know what you're solving. Be clear about your constraints (budget, staff capacity, data quality). Ask hard questions: What specifically will this tool do? What won't it do? What happens when it makes a mistake? What does implementation actually cost, and what's included? Can we start small and grow, or is it all or nothing?
Get references. Ask other nonprofits who've bought the tool whether it actually solved the problem they thought it would. Ask them what surprised them about the cost and the learning curve.
And remember: the vendor's success metric is you buying the tool. Your success metric is solving your actual problem. Those aren't always the same thing.
So What Now?
AI can be a good tool for nonprofits. Genuinely. But not for everyone, and not for every problem.
Start with clarity. Know what you're solving, whether you're ready to solve it, and who will manage the solution. Then, if the math makes sense and you have the bandwidth, add the tool.
If you skip those steps, you'll end up with another subscription you don't use and a team that learned, once again, that shiny tech isn't a shortcut to doing good work.
That's not an AI problem. That's a strategy problem. And that one's on you to solve.
This is what Coat Rack does, we help nonprofits think through their tech strategy before they make big decisions. If you're evaluating AI or any other technology, and you want someone to help you ask the right questions, let's talk.
