One Memory, Many Doors

Last July I published a UX manifesto with eleven principles. Principle 6 was "Modular, Not Monolithic," and the manifesto's one-line version of it read: "BeMo is a suite, but every piece stands on its own. Use one tool. Use them all. It just works." A month later I gave the principle its own essay. I have been going back through the record I put out while we were building BeMo, checking each claim against what building it actually taught. This edition is about the one where the instinct was right and the words were wrong, and about a framing I pitched in August that our own positioning rules now forbid.

What I claimed

The August essay made the case against the do-everything platform. Nonprofit software "tries to do everything" and delivers "bloated, slow, and overwhelming." Our answer was independence first: "Each tool works on its own. You can start with one. Add another later. No long onboarding. No platform lock-in." Connection was the bonus feature, described in exactly this much detail: "But when you do use both? They talk to each other behind the scenes. Seamlessly."

That was the whole architecture story. Independence was the product; connection was a nice thing that happened backstage.

What held

The anti-monolith instinct held completely, and it still shapes the product. BeMo never became the hundred-tab platform. It still does not do HR, finance, or program delivery, on purpose. And it is not a CRM, which people ask often enough that the scorecard should say it plainly: BeMo does not manage donor records or track interactions, and it is not trying to grow into the system that does. The question keeps coming because the category trained everyone to expect any nonprofit platform to eventually swallow every adjacent job. When one refuses, people check. There is still no configure-everything-before-you-see-value onboarding, and nobody has to adopt the whole thing to get use from one part of it. You can walk in through a single door.

The beta confirmed the entry half of the claim, too. People do come in for one thing: a funder search, a board communication problem, a pile of documents that needs to become an annual report. Nobody arrives wanting a platform. Score that part: held.

What the beta actually showed

Here is what the essay did not predict. In June, a beta tester who came in to find funders sent us written feedback, and almost none of it was about funder discovery. What she praised was that she could ask any question from anywhere: "Love that you can ask any question, no matter what the subject, even if you are in a particular section of BeMo." She praised the conversations that persisted. She praised the Knowledge Base telling her what it was missing after she loaded her documents.

She entered through one door and what she noticed was the building. Not one tool doing its job well, which is what I had advertised, but the fact that every room she walked into already knew what had happened in the others.

That is not what "modular" promised. Modularity, as I pitched it, made standing alone the value and connection the convenience. The beta says people experience it exactly backwards: the door they entered through is quickly taken for granted, and the thing they write to us about is the connectedness.

The word we banned

Now the confession inside the confession. This spring we wrote down the ways BeMo must never be described, an internal positioning canon that every piece of copy gets checked against. One of the entries: never call BeMo a "bundle" or a "suite of tools." The reasoning in that document is blunt: bundles imply independently conceived tools marketed together, and the products are expressions of one system, not separate tools packaged together.

The manifesto's own principle line, "BeMo is a suite," is now a forbidden sentence. I wrote the pitch that our rules now ban.

The deeper miss is in "they talk to each other behind the scenes." That sentence implies plumbing: separate systems with sync jobs between them, integration as a feature. What we actually built has no between. The apps read and write the same organizational memory. When your grant work shows up in your board report, nothing talked to anything; both surfaces were looking at the same knowledge. "Seamlessly" was doing a lot of work in that old sentence, and the honest version is that there were never seams to hide. Even that word is on our banned list now.

And "It just works" was borrowed confidence. In August 2025 nothing yet worked; the sentence was an aspiration wearing the grammar of a fact. The beta earned us the right to describe specific behaviors instead: the system says so when it does not know something, and it tells you what is missing. Those are claims I can stand behind because testers reported them, unprompted.

What changed in BeMo

The vocabulary went first: "suite" is banned, and we describe BeMo as one platform with several ways in. The messaging followed the beta's lesson: meet people at the pain point that brought them, and let the coherence reveal itself instead of being announced. And the architecture claim got honest: not tools that integrate, one memory with many doors.

In one sentence, the delta: I pitched tools that stand alone and also connect; building BeMo taught us that the connection is the product, and standing alone is just the door you walk in through.

Then: use one tool, use them all, it just works. Now: enter anywhere, and everything you do lands in one memory that every other surface can see. What we got right was refusing the monolith. What we got wrong was thinking the alternative was a suite.