Your nonprofit can build its own software now
If you know your problem and have your data, AI can build the tool nobody was ever going to sell you. Some of them you'll use once and throw away.
Can a nonprofit really build its own software with AI?
Yes. You describe the tool you need in plain language, and AI writes the code, builds it and helps you fix what's off. People call this vibe coding. You won't write a line of code, and you don't need a developer on staff to get something useful running in an afternoon.
Scott Williams · Recorded conversation, Sep 14, 2026you don't actually code when you vibe code. You just talk to your computer. That's literally what it is.
I'm not a web developer. I've run an agency for 11 years, so I know how to talk about this stuff, and I still couldn't build a website by hand. In October I rebuilt the If Possible website overnight anyway. I spent a few days, 20 minutes at a time, getting our design system exactly right. Then I asked Claude to build the whole site while I slept. I woke up to a preview of the entire website and had maybe 10 or 15 notes for it.
What made that possible was knowing the concepts. In the age of AI, you mostly need to know how things work and what you want.
What do you need to build your own nonprofit software?
You need a problem you can describe and the data that goes with it. That's the whole list. Give AI both, ask it to help, and it will propose a tool, build it and adjust it as you react.
I said this to Colleen in September, right after showing her a personal command center I'd built:
Scott Williams · Recorded conversation, Sep 15, 2026You just need to know your problems. And you need to have data. That's all. And all you do is you just give it your data, and you give it your problems, and you say fix it.
In practice, that can be as simple as gathering a spreadsheet and a paragraph about what's frustrating you, putting both into Claude and asking for a live artifact, a small interactive app that runs right in the chat. If you're not sure where your problems live, open your spreadsheets. Ask why you're tracking each column and what decision it's supposed to help you make. The answers point straight at the tools you're missing.
What software should a nonprofit build first?
Build the tool for a problem so specific to your organization that no company will ever make it. Those are the problems commercial software skips, because a market of one isn't worth selling to. For you, it might be worth hours every week.
Colleen asked this in September, and it's still my favorite question we teach:
Colleen Cook · Recorded conversation, Sep 14, 2026what is the hyper specific problem to your organization that nobody's ever going to make a market solution for because there's no profit in it. But you actually do need it.
Here are a few we've built for ourselves at If Possible:
| What we needed | What we built | What it replaced |
|---|---|---|
| A way for clients to comment on website pages | Our own commenting tool | A monthly software subscription |
| One place to review new videos | A review app where approving a video sends it on to be posted | Clicking through several tools by hand |
| A view of tasks, calendar and ideas from every tool | A personal command center | Checking each app one at a time |
| Four DVDs burned for a visitor | A burner app Claude built, used once and deleted | Hunting for a serial number |
None of these would ever be sold to us as a product. Each one took a conversation, plus some back and forth.
What is disposable software?
Disposable software is a tool you build for a single job and delete when the job is done. Building used to cost so much that every piece of software had to be bought, kept and maintained for years. With AI, a tool can take five minutes to make, so it can also be thrown away.
I learned this from a DVD. I needed to burn four of them for a visitor, and I didn't have the serial number for the burner software. So I asked Claude for help.
Scott Williams · Recorded conversation, Oct 22, 2026give me about five minutes. I'm gonna build software to burn your DVDs. And then when we're done with burning these four DVDs, I'm gonna throw the software away.
That's exactly what happened. It built the software, burned the DVDs and deleted the software. Once you've seen that, every annoying task looks different. Now the first thing I ask is whether AI could make me a tool for this one job.
Does homemade software have to be good enough to sell?
No. It only has to solve your problem. Commercial software has to work for thousands of customers. Yours has to work for your team on Tuesday. That's a much lower bar, and it's why a tool you build in an afternoon can beat one you pay for every month.
it can solve a problem for you, it's good enough. You don't have to take it to the market.
One caution from experience. Build as many front ends as you like, and keep your data in one home. If you don't like how a tool looks, build another one on top of the same database. Your information stays organized, and the tools on top of it can come and go.
Should a nonprofit replace the software it pays for?
Sometimes, and it's worth asking every time a renewal comes up. If you use a small slice of an expensive tool, AI may be able to build that slice for you. Keep paying for systems that hold critical records, handle money or need real security.
Colleen asked a cohort member in October why she'd keep using a CRM that was only sort of working when she could build exactly what she needed in about 20 minutes. That question is the habit I want every nonprofit leader to pick up. Before you buy, ask whether you could build it.
Here's a starting prompt for your first tool:
``` Here is a problem my nonprofit has: [describe it in a few sentences]. Here is the data that goes with it: [paste or attach the spreadsheet]. Can you help? I'd like a live artifact I can click through, built just for this. Ask me questions before you build if anything is unclear. ```
Expect to react to the first version, ask for changes and try again. That back and forth is the build.
If you're weighing what to buy versus what to build, read tools vs systems vs coaching next. Your first 30 days with AI shows where a first build fits in a rollout, and there's more like this in our implementation articles.
For one practical idea each week, join the If Possible newsletter. If you'd like to build your first tools with us and other nonprofit teams, get on the cohort waitlist.
Frequently asked questions
Can a nonprofit build its own software with AI?
Yes. Describe the problem and share the data that goes with it, and AI tools like Claude can build a working app, often in an afternoon. You don't need to know how to code, and the tool only has to work for your team.
What is vibe coding?
Vibe coding means building software by describing what you want in plain language and letting AI write the code. You react to what it builds, ask for changes and keep going until the tool does what you need.
Is it safe to build our own nonprofit software?
For internal tools that organize work, yes, with care. Keep sensitive records, payments and donor databases in established systems with real security, and use homemade tools on top of them or for tasks that don't touch protected data.
What should our first homemade tool be?
Pick a problem specific to your organization that no vendor would ever build for, like a report you assemble by hand every month. Gather the spreadsheet, describe the frustration and ask AI for a live artifact.
Know an executive director who needs this?
