Our money has a copilot — blog.stancs.com

The health entry described Claude as a trainer who never clocks out. This is the same idea pointed at money — and honestly, it might be the more consequential of the two. I won’t share specifics or numbers here, but the system is the story, and the system is worth telling.

Fully informed, by design

You can’t get real financial guidance from something that only sees fragments. So we went the other way: Claude has access to our complete financial picture. Every year of tax returns we have. Every transaction across every account, with near-real-time history pulling from Monarch — refreshed at least every few days. And when a question needs source data, I’ll log us into an investment site or an HR system so it can pull the actual statement or the actual paycheck detail, straight from the record.

All of it lives locally, as files in our own workspace, and works as context. The result is that I can talk about money exactly the way I talk about training and nutrition: any question, anytime, grounded in our actual situation rather than a calculator’s assumptions.

It found money we were leaving on the table

Here’s the concrete proof it works. For the first time in many years, we enrolled in account types we’d never touched — HSAs among them — because Claude noticed we weren’t using everything our employers offer, laid out how the benefits actually work, what they’re worth, and where the risks and catches are. The enrollment wasn’t a leap of faith; it was the endpoint of weeks of back-and-forth questions until we understood the thing well enough to be confident. That’s the difference between advice and understanding — the house rule here is that Claude explains and models but doesn’t tell us what to invest in. Understanding turned out to be enough.

It also runs a standing sweep in the other direction: every few weeks it combs the transaction history against our employer perks and hands me a list of expenses — health, fitness, and the like — that qualify for reimbursement. Money that used to evaporate on a receipt now comes back, because something with perfect recall reads every line item.

One more trap it watches: a mid-year job change in the family meant a retirement contribution limit split across two payroll systems that can’t see each other. Claude runs the combined math against real pay stubs, projects to year-end, and tells us when to nudge the percentage — every other Thursday, on schedule.

The weekly what-if machine

The most important part isn’t tracking — it’s modeling. Our finances don’t change often, but our questions never stop, and roughly weekly we sit down and run scenarios:

What does a move look like? A location change, a job change, a promotion — or the other direction?
What if one of us loses a job? What if we both do — how long are we fine, and what changes first?
What are we actually on track to leave our daughter?
And the un-plannable stuff — the air conditioner that dies on a Tuesday — gets folded into the model the week it happens, with suggestions for how to rebalance around it.

Every scenario runs against the real numbers, so the answers aren’t platitudes — they’re projections with our name on them.

The forest from the trees

Here’s the part that surprised me: the biggest return hasn’t been financial, it’s been psychological. I have a tendency to fixate — on carrying any debt at all, on the transaction-level noise. The weekly review keeps handing me the long view instead: the direction of the whole picture, over years, with every what-if already stress-tested. It’s hard to overstate what that does for your peace of mind. Not because the news is always good — because the picture is always whole.

Same lesson as the health tracker, one level up: get the ground truth into a place you control, let something tireless do the arithmetic, and spend your own attention on the decisions — and on sleeping at night.