Hey there, fellow accidental techie! π
Reader, your executive director forwarded you an article last Tuesday. Subject line: "We need an AI strategy." The article attached was 2,400 words long. By paragraph three you were already behind. By paragraph seven, you'd been informally volunteered to explain what a large language model is at the next all-staff.
You're the accidental techie. You've kept your org's CRM running without formal training, survived the move from Dropbox to Google Drive, and held together at least one tech migration that should have gone sideways. You've been doing this for years. And now, apparently, you need an opinion on artificial intelligence.
Here's what's worth saying before anything else: most of the people telling you to "adopt AI" have no clearer picture of what that means than you do.
What's in this issue:
β
What the AI adoption numbers actually say β and why they should make you feel better, not worse
β
The vendor noise problem and how to cut through it
β
What I observed working alongside nonprofit tech leaders for years
β
What a real 30-day pilot looks like at your scale
β
The one question to answer before you touch any tool
β
Three moves to start this week
β
The Intentional Techie July cohort
β
What I'm reading
Before we get into it β one thing this week.
A fews back, I ran a live session walking through all four bridges together β the full framework and real examples. If you were there, thank you. If you missed it, the recording is available now.
This is the clearest version of everything I've been building toward in this series. If you're going to watch one thing before enrollment closes for the July cohort, make it this.
|
|
From Accidental to Intentional: The Four Bridges Every Nonprofit Tech Leader Needs to Cross
|
What the numbers actually say
By 2026, 92% of nonprofits are using AI tools in some capacity. That number sounds like momentum until you read the follow-up: only 7% report major improvements in their organizational capability. And 76% have no formal AI policy of any kind.
What those numbers describe isn't a sector that's figured this out. It's a sector where almost everyone is doing something, and almost no one is doing it on purpose.
That gap is exactly where most accidental techies are sitting right now. It's not because your org is behind. It's because the tools arrived faster than any sensible framework for using them.
The vendor side is making this harder. Every piece of software your org uses now has some version of an "AI" badge on it. Your donor database is AI-powered. Your email platform has AI-assisted subject lines. Your project management tool just added an AI assistant. Most of this is existing functionality with new branding, and sorting through it while also doing your actual job is exhausting in a way that's hard to explain to people who don't do what you do.
What I saw in years of working alongside nonprofit tech leaders
I spent years at the Technology Association of Grantmakers sitting alongside nonprofit technology leaders from organizations of every size and budget. Every major tech cycle follows the same arc: first comes the panic, then the pivot, then the plateau, where people figure out what actually sticks and quietly drop the rest. I've seen it with cloud migration, with mobile-first mandates, with chatbots, with blockchain. AI is moving through that arc faster than anything I've witnessed before. But the arc is the same.
The organizations that come out ahead aren't the ones who adopted first. They're the ones who adopted with a clear reason.
So what does that look like for a 12-person org where "IT staff" means you?
A real pilot doesn't have to be complicated. Pick one task your team does at least weekly, something that follows a repeatable pattern and doesn't touch sensitive data. Meeting summaries. First drafts of policy language. Social media copy variations from a post you've already written. Use one free tool for 30 to 60 days. Measure what changes. A tool like Claude, available through an interface like Cowork, can take a rough brain dump and give you structured draft content in minutes. It won't be polished. It's a starting point, and that's the whole point.
The one question to answer before you touch any tool
Your organization holds data that belongs to real people. Donor financial information. Client contact records. Program beneficiary data. Sometimes health information or immigration status. Consumer-grade AI tools, including the free tiers of the most popular platforms, may use your inputs to train future models. Some tools store conversation history indefinitely. Others are explicit about data handling. Many are not.
This isn't a reason to avoid AI. It's a reason to know exactly what you're handing over before you hand it over. Using an AI tool to draft grant narrative language is completely different from using that same tool to process case notes on the people your organization serves. The first is low-risk. The second is not, and the line between them can blur faster than you'd expect when a deadline is close and a free tool is right there.
Where does AI actually help right now, at your scale? Quite a bit, in the right places. Grant writing is the most common use case in the sector for good reason: the output is a draft, not a submission, and an imperfect first pass costs you nothing but time. Meeting summaries save real time, especially when a staff member missed a session. Communication templates, onboarding documentation, training outlines, policy drafts: these all work well because they're starting points. AI is a capable drafting assistant with no understanding of your organization's history or relationships. It doesn't know why your last ED left, or how your biggest donor prefers to be thanked, or what the board got nervous about two years ago. Use it where that context doesn't matter, and keep it away from the places where it does.
Join The Accidental Techie Community in Linkedin
|
Three moves you can make this week
Audit what data your org actually holds. Before you touch any tool, spend 20 minutes on a list. Write down every category of sensitive data your organization holds. Donor financial information. Client records. Program participant health data. Immigration status. Anything that, if it ended up somewhere it shouldn't, would be a real problem for the people who trusted you with it. That list is the foundation of your AI policy, even if you don't call it that yet. It also gives you your first clear answer when someone asks what guardrails you have in place.
β
βPick one task and run a real 30-day pilot. Not three tasks. One. At the end of 30 days, ask two questions: Did this actually save time in a way I'd notice if it stopped? And would I be comfortable telling our ED exactly what data I put in to make it work? If both answers are yes, you have something worth keeping. If either answer is no, you've spent 30 days on a free experiment that gave you real information. That's useful. It's faster than most orgs move.
β
βBuild a two-question filter for every future AI pitch. There will be many more pitches. Run these two questions on every one of them: "What data does this tool train on, and where does it go?" And: "Could I explain how we're using this to our board without anyone getting nervous?" If a vendor can't answer the first question directly, you have your answer. If your honest reply to the second is "probably not," stop there.
β
βSuccess looks like this: the next time your ED forwards an AI article, you have a calm, one-paragraph response about what your org is already doing, what guardrails you've put in place, and what you're still evaluating. Not because you've solved artificial intelligence. Because you started on purpose instead of by accident.
β
β
|
|
Join the July Cohort
Sign up for the Waitlist to receive the annoucement
|
Until next time,
Hugo
Join the Waitlist
A 6-week cohort program for nonprofit professionals ready to stop firefighting and start shaping technology strategy.
Join the waitlist for early access and beta pricing.
|
β