Most mornings I check the Lucie Money waitlist number. It’s usually the same as the day before, or up by one. There’s a reason for that: we don’t have anything to show yet.
From the outside, building a startup can sound cool: demos, pitch decks, presentations, the launch, charts (going up and to the right, of course). But at the moment, almost none of what we actually do looks like that. Day after day, our current work is building the stuff that nobody sees: the foundations.
None of this is unique to us, of course. I imagine every complex business is mostly foundations. The bit you see (the shopfront, the app, the brand) sits on top of a lot of unglamorous work no customer ever notices, and getting that work right is most of what decides whether the business is any good. What’s different for us at the moment is the degree. At this point in time there is no shopfront. Lucie is, give or take, entirely invisible. There’s only the foundations.
And for a money product, the foundations are most of the point. The one thing a tool that handles your money can’t do is be wrong. Get a number wrong and everything built on it is wrong too: conclusions, advice, trust, eventually everything. On top of that sits a thicket of regulation that exists precisely to make businesses get this right. So the invisible work isn’t a phase we’re pushing through to reach the good bit. It’s the heart and soul.
Let me mention a few examples.
How should Lucie show you a single transaction, and a single account? You’d think that was a solved problem. It isn’t. We started working deeply on how an agentic solution should use and present financial data over a year ago, and we’re still not finished.
We first had to decide how to ingest and process financial data to keep it accurate and verifiable but still operate within the context of an agentic overlay that needs access to that data for reasoning. This involved an explosion of issues across every facet of LLM orchestration and verified data pipeline management, which James, our CTO, has led.
Only once that architecture was established could we move on to layout and data presentation UX, which raised its own host of questions. And the moment we did, the detail exploded once again. Where does each piece of data even come from? The open-banking standard offers multiple options for ingesting any one data item. And our CDR partner provides its own overlay. A single foreign-currency purchase comes with an original amount, a converted amount and an exchange rate, none of which turn up in a tidy field; they have to be dug out of a free-text description that changes from one bank to the next. Then come the permutations: different payment types, transaction types, account types, each with its own quirks… don’t get me started.
This one apparently small thing, showing you your money clearly, will take about a year all up. A user will glance at it for half a second and move on.
Then there is the regulatory framework. An app that helps people with their money sits under an intricate body of regulation, and working out exactly where those rules apply, and what we need to do about it, is a substantial piece of work, with particular aspects unique to agentic systems that give personal financial guidance. How this eventually applies is yet to be settled and may take considerable time, resource and cost, and we’re only at the beginning of the process. But none of this is visible to anyone outside, and all of it has to be right.
And lastly there’s risk. Building institution-grade risk discipline into a startup sounds faintly absurd when you say it out loud, and it’s certainly unglamorous. But with money there’s no version where you do this later. It has to be there from the start, for the legitimacy and security of everything and because the regulation requires it.
Which brings me back to the waitlist number. The awkward part of this stage is that the most important work is the least visible, and there’s not much I can do about that. Nobody’s going to get excited about the correct rendering of a foreign-currency transaction; I can’t show potential customers a risk register and control catalogue.
It might be tempting to rush something, anything, to the surface, just so there’s a thing to point at. But decades of working in banking and technology tell me that’s the wrong instinct. The true discipline is to do the invisible work properly while it’s invisible, so that the surface manifestations will have a solid foundation on which to stand.
We’re in the middle of this, so take all of this as what we’re doing, not what we’ve done. But I’m increasingly convinced that the stuff nobody sees is the stuff that decides whether the rest is any good. So that’s what we’re doing.
If Lucie Money sounds like something you’d want, please subscribe to the waitlist at lucie.money. It might even tick up by one.