On October 1, a chatbot humbled me in front of hundreds of people.
Hundreds of fundraising professionals had signed up for our Agentic Giving webinar to see what happens when a donor asks an AI assistant where to give and the assistant actually helps them do it. My job was the fun part: the live demo. Ask the assistant about a cause, watch it find a real nonprofit, pick the right campaign, and complete a gift right there in the chat.
I had a plan. I had rehearsed it in a dry run. I had tested my prompts that morning like a kid checking his backpack for the third time before the first day of school.
Everything worked right up until the payment. Claude handed me off to Stripe Link, I approved the gift with my saved payment method, and then nothing happened. Fundraise Up couldn’t detect that I had approved it, so as far as our system knew, I was still sitting there deciding. Claude did what any polite assistant does when it gets confused: it produced an error message and tried to reroute me to a fallback link.
There is a special kind of silence on a webinar. On a stage, you can at least see the audience wince. On Zoom, you get a chat box, a ticking clock, and the sound of your own voice saying, "let’s give that one more try."
A couple of weeks later, I’ve decided I’m glad it happened, for a few reasons.
Live demos of new technology fail
In 1998, Bill Gates was showing off Windows 98 in front of a packed room when the computer blue-screened on stage. He waited a beat and said, "That must be why we’re not shipping Windows 98 yet." Steve Jobs’s first iPhone demo only worked because his engineers mapped out a single "golden path" through the prototype, and he had to follow it exactly or the phone would freeze.
I’m in no danger of being confused with either of those guys. Still, if Microsoft and Apple couldn’t make brand-new technology behave on command, a customer success manager with a laptop and a dream wasn’t going to bat a thousand either.
The plumbing broke
Agentic giving depends on several systems shaking hands in real time: the AI assistant, the nonprofit’s public information, the donation platform, and the payment layer. Each of those pieces is changing week to week. AI assistants decide on their own schedule when to read a website, and the same question can get a different answer on Tuesday than it did on Monday. No assistant guarantees a recommendation to anyone, and we try to be very up front about that.
When one handshake slips, the whole demo slips, and ours slipped at the payment step.
Everyone saw where the technology really is
We could have played a polished recording. It would have looked great, and every nonprofit watching would have gone home believing this works flawlessly today. Then they’d turn it on, ask ChatGPT about themselves, and wonder what they did wrong.
Instead, everyone on that call saw where the technology actually is: real, working most of the time, and still rough around the edges. That’s more useful to a fundraiser deciding whether to turn it on.
The donors are moving anyway
Blue State’s 2025 research found that AI-referred traffic to nonprofit websites grew roughly 1,000% year over year, and donors who arrive from AI give larger average gifts ($240 vs. $210). It’s still a small slice of total traffic. But that slice is your donors asking a machine where their money should go.
And it’s growing fast. Since we released Agentic Giving on September 1, agent traffic to Fundraise Up checkouts has grown 3,000%, helped along by new agents like Meta’s Muse and OpenAI’s Dots. We don’t have every i dotted and t crossed yet. The payment handoff that tripped me up on the webinar is exactly the kind of thing we’re working through right now, while the volume is still small enough to learn from. That’s the whole point of preparing early. When donors start expecting AI giving to work, we’ll be ready.
Want to know how AI assistants describe your organization today? Run the free AI Visibility Assessment. You can watch the full webinar, glitch included, here.
AI Visibility Assessment
Wil is a principal success manager at Fundraise Up who lives in New York. He has been with FRU since 2021 and is one of our longest tenured employees. After 6 years, he pretty much has a PhD in Fundraise Up and has loved being a small part of hundreds of nonprofit teams. He previously worked at a nonprofit in Knoxville, TN and an AWS cloud computing consultancy and loves to merge his charity and technology expertise to serve organizations.