Until the Popcorn Stops Popping

In February, when we were planning the beta, I asked my co-founder and software engineer, Lee, how long it would run. Two or three months, he said. Then he corrected himself. "Until the popcorn stops popping."

Lee has a way of explaining things through analogies, and I have worked beside him long enough to know I would eventually understand what he meant. We opened the beta in May. It is the end of September, and the popcorn is still popping.

Four months in, I can tell you what the popping has taught us.

Where we started

Before the first organization logged into BeMo, we spent more than a year and a half on calls with executive directors, development directors, consultants, and founders, including people preparing to hand off organizations they had spent decades building. We talked about workstreams, tools and pain points. We called it Project Talk to People because that is what it was. Almost none of them described a technology problem. They described a person holding too much, and nothing built to hold it for them.

Two things came out of those conversations that we embedded into BeMo. The first is that nobody in this sector owes a piece of software their trust. Most of the people I talked to had a story about the platform that was going to change everything. Usually the software was fine. The promise is what broke. The second is that if a machine does work on your behalf, you should be able to see the work. What it used, what it ignored, where it was guessing. We started calling that "showing the receipts," and every design decision since has had to pass it.

The rule under every feature

"Human in the loop" is a phrase everyone uses and very few people explain. Here is what it means at our company.

BeMo proposes. A person confirms. Nothing becomes true in your organization's record until someone with a name has said yes to it. BeMo can draft a grant narrative from what it knows about your organization and the funder, show you where the information came from, and wait. You change it, or you do not. The version that lands is the one you approved, with your name on it. We say it in three words:

You hold the pen.

The people we talked to were not asking for a machine to decide for them. They were asking for less to hold, and those are different asks. The receipts and the pen are the same principle from two sides. You can see the work, and you decide what becomes true. That is what trust means to us, and it is the test every feature has to pass before it reaches you.

What four months of popping taught us

Lee and I use BeMo on our own work first. You have to notice what feels intuitive and what feels clumsy, find the things that break in real time, then go back, fix them and try again. A feature is not ready because it works once. It is ready when it reliably hands the pen back. Only then does it go to the people who have agreed to break it for us. Here is some of what they found.

"Done" is not done until someone checks. In August, a user and BeMo agreed on a wording change, BeMo confirmed it, and the document did not change. She invented her own workaround before we knew it was happening: always ask it to check. The rule we wrote that week is that BeMo reads the document back before it says anything is finished. She caught it happening again in September, which told us that writing the rule was not the same as making the behavior dependable.

Nothing you send us is in the wrong place. Through the summer, a user's emails arrived with subject lines asking whether the problem was ours or hers. It was usually ours. Sometimes it was ours in a way that made it look like hers, which is the worst kind.

People can only try what they can imagine. In August, a user asked whether we would gather the early users so they could show each other what they had made, because she suspected someone else was doing things she would never think to do. She was naming a ceiling we had not seen. The guide we wrote for new signups starts from that: real scenarios, each beginning with something the organization already has, so people recognize themselves in a few of them instead of staring at a blank page.

Where we are today

Anyone can open an account now, free. What is inside has been through the loop above: the organization's own record, a conversation grounded in that record, tools for creating and working with documents, funder discovery built from what funders actually did, and teaching that arrives inside the task. Different tools, one rule underneath them: show the work, and leave the decision with the person using it.

The signals that tell us we are on the right path are not the loud ones. The leader of a library friends group ran her annual report in BeMo with almost no help from us. It drafted from what it already knew about the library, she decided what belonged in front of donors and what stayed internal, and the report that went out was hers. She has since brought BeMo to a coffee talk with a dozen local nonprofits. And the person who reports the most problems is also the person who sends the most notes about what worked. Both halves, unasked, in the same tone.

That is what building trust in beta looks like.

Why the popcorn is still popping

What we can use ourselves, we test ourselves. Then we put it in front of the people who have agreed to push it harder than we can. What survives that loop moves forward. What still surprises us stays with us until it is dependable enough to stop.

So the beta is still open, and the invitation is the same one we made in June. Come find the edge with us. Trust does not begin after all the rough edges are gone. It begins while you can still hear the popcorn, with the pen in your hand.