It's starting to feel like a product

It's starting to feel like a product

A month ago I wrote about opening the Lucie Money app and watching it work for the first time. I was able to see a list of accounts, check balances and view recent transactions.

This week I was able to see more: multiple accounts, full transaction lists dynamically filtered, account details views showing balance details, interest rates and more.

It’s starting to feel like a product.

What’s there now

The set of widgets for accounts and transactions is now pretty-much built out. A user can see all their accounts (or a selection of them) in one place; open any one of them for details; and open any transaction for the full detail behind it.

Account list
Lucie account list showing four accounts with their balances and a combined credit, debit and total position
Balance
Balance card for an everyday access account showing the current and available balance
Interest rate
Interest rate card for a savings account showing the standard rate, the bonus rate and the conditions for earning it
Payment due
Payment card for a credit card account showing the closing balance, the due date and the minimum payment
Recent transactions
Transaction list for an everyday access account showing five recent transactions with merchant names, categories, dates and amounts
Transaction details
Transaction detail card showing a single purchase with its amount, time, account and category

The main thing to emphasise is that every figure in these widgets comes straight from the user’s bank, not from Lucie’s AI. In Lucie, the AI’s job is to explain what’s going on with the user’s money. That key architectural construct is where most of the effort has gone, and it’s why this stage has taken so long.

None of this is exotic. Every banking app on your phone does some version of it. But we had to create all the foundational capability first before we could build the fancy stuff. And we had to build ours from nothing, mind you.

Test data is too well behaved

The next real step is getting off test data. Test accounts are tidy with sensible names, sensible amounts, sensible dates, and nothing missing. Real accounts are not like that, and we don’t yet know all the ways they’ll trip us up.

Some of it we can guess at already. Merchant names can be inconsistent between banks; we’ve only got a handful of merchant logos so far (you need hundreds before a transaction list starts looking like places you’ve actually been).

Most of it, though, we’ll only find out once real data starts arriving. That’s just how it goes.

What comes next

First, real bank data. We’re at the contract stage with our open banking partner and working through the last of the legal detail. Once that’s settled, Lucie can start connecting to real accounts.

Then the part people ask about most: making sense of spending. Sorting transactions into categories, working out what’s normal for each person, spotting when something isn’t, and then saying something useful about it. That’s where it starts to get interesting.

Subject to how our regulatory work goes, we’re still aiming to get preliminary functionality out in the second half of this year.

So there’s still a lot to do and in some ways we’re just at the beginning. But a month ago I couldn’t have opened an account and looked at what was behind a transaction. This week I could. That’s progress.

There’s a waitlist at lucie.money.