AI is getting a lot of attention right now, but for an understaffed NonProfit, the most important question isn’t whether the organization is “using AI.”
It’s whether AI can give your people some time back.
That’s a much more practical place to start.
If an employee spends three hours every week turning notes into reports, organizing information, drafting routine communications, or doing other repetitive administrative work, there may be an opportunity.
If an approved AI tool can safely reduce three hours of work to one, that’s two hours of capacity returned to the organization every week.
Multiply that across a year.
Then multiply it across several employees.
Now AI has a business case.
On the other hand, if your NonProfit signs up for another AI tool because everyone is talking about it, but employees barely use it—or spend almost as much time correcting its work as they did doing the work themselves—you haven’t gained much.
You’ve probably just added another subscription, another login, another tool to manage, and another place where organizational information could end up.
AI isn’t the strategy. Getting useful staff time back is the strategy.
That distinction matters.
AI Can Be Very Good at Getting You Past the Blank Page
One of the most practical ways employees can use AI is also one of the least exciting.
Creating first drafts.
Imagine your program director needs to write an update about a new initiative.
She knows the program inside and out, but she's staring at a blank Word document trying to figure out how to organize everything she wants to say.
She might spend 45 minutes getting the first few paragraphs into reasonable shape.
AI can potentially change that process.
Instead of starting with a blank page, she can begin with appropriate, non-sensitive information and ask an approved AI tool to help organize her thoughts into a first draft.
Now she isn't asking AI to do her job.
She's giving herself something to react to.
She can correct it, rewrite it, add the organization's voice, remove things she doesn't like, and make sure the final version accurately reflects what she wants to communicate.
That distinction matters.
AI doesn't necessarily need to produce the finished work to save time.
Sometimes its biggest value is helping an employee get from nothing to something.
Meeting Follow-Up Can Consume More Time Than the Meeting
Here's another common example.
Your leadership team has a one-hour meeting.
The meeting ends.
But the work created by that meeting doesn't.
Someone has to organize the notes, identify action items, summarize decisions, create follow-up tasks, and perhaps prepare a recap for people who weren't there.
Suddenly the one-hour meeting has generated another hour of administrative work.
Multiply that by several recurring meetings every week.
With the right process and appropriate safeguards, AI may be able to help turn approved meeting notes into a structured summary, identify action items, organize decisions, or create a first draft of follow-up communication.
Again, the goal isn't to remove human judgment.
Someone still needs to review the result.
The goal is to reduce the amount of time a capable employee spends doing repetitive administrative work after the actual thinking has already happened.
If AI turns 45 minutes of post-meeting administration into 15 minutes, you've just returned half an hour to that employee.
Do that twice a week and you've returned roughly 50 hours of staff capacity over the course of a year.
Now we're talking about an outcome that matters.
AI Can Organize Information, Not Just Write Things
ChatGPT and other AI tools often get described as writing tools.
That’s only part of the opportunity.
A lot of staff time is spent simply trying to make sense of information.
Imagine your leadership team finishes a planning session with six pages of notes.
Someone now has to read everything, identify common themes, group similar ideas, create a summary, and prepare something useful for the next meeting.
With appropriate information, AI may be able to help with that first pass.
The same applies to research, long documents, brainstorming sessions, lists, surveys, or other information that needs to be categorized or summarized.
Again, the employee still needs to review the result.
AI can misunderstand context.
It can omit something important.
And it can produce an answer that sounds remarkably confident while being wrong.
That last one is particularly important.
A polished answer and an accurate answer are not necessarily the same thing.
Your people still need to know the difference.
Human Review Isn’t Optional
There’s a temptation with automation to look only at the time it takes to generate something.
A task used to take an hour.
ChatGPT generated it in four minutes.
We saved 56 minutes.
Maybe.
How much time did someone spend checking it?
Did they have to rewrite half of it?
Were the facts correct?
Did it accurately reflect your organization?
Would you be comfortable sending it to a funder, donor, board member, employee, client, or the public?
Those questions are part of the calculation.
If AI generates a report in five minutes but an employee needs 45 minutes to verify and repair it, you didn’t save 55 minutes.
You saved ten.
Ten minutes might still be worthwhile, but leadership should understand the real return rather than the theoretical one.
That’s why measuring AI productivity matters.
Measure the whole process, not just how fast the tool produces an answer.
Be Careful What You Trade for Efficiency
There’s another part of the equation.
Data.
Health and social-services NonProfits may handle a significant amount of sensitive information: client records, employee information, health-related information, financial information, donor information, case notes, credentials, internal documents, and other data that has been entrusted to the organization.
Employees shouldn’t copy that information into ChatGPT or another AI tool simply because doing so makes a task faster.
This is where organizations need practical rules.
Which AI tools are approved?
What information can employees use with them?
What information cannot be entered?
Can employees use personal ChatGPT accounts for organizational work?
Who evaluates a new AI application before an employee starts using it?
Who does an employee ask when they aren’t sure?
These shouldn’t be trick questions.
The easier the rules are to understand, the easier they are for employees to follow.
Don’t Turn AI Into Another Collection of Software Nobody Owns
There’s an AI tool for just about everything now.
One writes emails.
One takes meeting notes.
One creates presentations.
One summarizes documents.
One creates images.
One analyzes spreadsheets.
And another promises to automate practically everything.
It’s easy to see where this goes.
Development signs up for one.
Operations buys another.
Someone in administration starts paying for ChatGPT personally.
A program manager finds a meeting assistant they like.
Six months later, the organization has five or six AI applications, multiple subscriptions, organizational information scattered among them, and nobody really knows what everyone is using.
That isn’t efficiency.
It’s technology sprawl with an AI label on it.
Before adding another tool, find out whether something you already own can solve the problem.
For NonProfits using Microsoft 365, that may include understanding where Microsoft Copilot fits and what capabilities may already exist within the Microsoft environment.
Other applications your organization already uses may also be adding AI features.
Fewer well-managed tools are generally easier for employees to learn, easier for the organization to support, and easier to secure than a collection of applications adopted independently by different departments.
Don’t Create an Unofficial AI Department Either
There’s another predictable way this can develop.
One employee gets interested in ChatGPT.
They learn how to use it well.
Other employees notice.
Soon that person starts getting questions.
“Can ChatGPT do this?”
“Which AI tool should I use?”
“Is it okay if I put this document into it?”
“Can you help me write a prompt?”
“Should we buy this AI application?”
The employee was trying to save time.
Now helping everyone else with AI has become another part of their job.
This is the same problem many growing NonProfits already have with IT.
Someone is “good with computers,” so technology gradually becomes their second job.
There’s no reason to repeat that pattern with AI.
It’s helpful to have employees who are interested in new technology and willing to experiment with it.
But decisions about approved tools, security, information handling, licensing, and organizational standards need clear ownership.
Otherwise, AI creates another layer of work instead of removing one.
Measure AI in Hours, Not Hype
For an understaffed NonProfit, there’s a fairly simple way to judge whether an AI use case is worthwhile.
Start with the current process.
How long does it take?
Then test the new process.
How long does it take including human review and correction?
If preparing a weekly report normally takes two hours and an AI-assisted process brings it down to 45 minutes, you have something worth paying attention to.
That’s 65 hours of capacity over a year from one weekly process.
Now ask what that employee can accomplish with those 65 hours.
That’s the outcome that matters.
Maybe it means more time with clients.
More attention to grants.
More time developing programs.
More donor communication.
More time managing employees.
Or simply enough room in the week for someone to get through the responsibilities they already have.
That’s a much better measure of AI success than how many employees have ChatGPT accounts.
Start With One Problem
There’s no need to turn this into a massive AI initiative.
Pick one problem.
Preferably an annoying one.
Find something repetitive that employees already know consumes too much time.
Understand the current process.
Make sure the information involved is appropriate for the AI tool you’re considering.
Test whether AI can make the process faster.
Have a human review the result.
Measure the actual time saved.
If it works, document what you did and repeat it.
If it doesn’t, move on.
That’s still useful information.
Not every process needs AI.
Not every problem is a technology problem.
And not every shiny new tool needs a place in your organization.
AI Should Give Your NonProfit Capacity Back
NonProfits don’t need AI for the sake of having AI. They need more capacity. That’s the opportunity worth paying attention to.
If ChatGPT, Microsoft Copilot, or another approved AI tool can safely take repetitive administrative work off an employee’s plate, there may be real value there. But the process should start with the work, not the software.
At I-M Technology, we look at technology through that lens.
What problem are we trying to solve? What does a successful outcome look like? Does the technology actually move the organization toward that outcome?
With AI, one of those outcomes should be easy to understand:
Your employees get some of their time back. That gives them more capacity for the programs, people, fundraising, leadership, and day-to-day work your NonProfit exists to do.
If AI accomplishes that safely and measurably, it’s doing something useful. If it doesn’t, adding more AI probably isn’t the answer.
