The family office reporting stack, and where it breaks
Most family offices run on a spreadsheet, a folder of PDFs and one person who understands both. Here is where that arrangement actually fails.
Ask a family office what they use to track the portfolio and the honest answer is usually a workbook. Sometimes a very good workbook, built over years, with the quirks of every fund encoded in it.
It is easy to be sniffy about that. It is also wrong: the spreadsheet is there because it is the only tool flexible enough to model what an actual family office holds. Fund commitments, a few SPVs, some direct positions, a property, a loan to a portfolio company, three entities and a trust. No off-the-shelf product handled all of that, so someone built it.
The workbook is not the problem. The problem is what surrounds it.
Where it actually breaks
It breaks at the entity boundary. One workbook per entity is manageable. The consolidated view across entities is a fourth workbook that reads from the other three, and it is always slightly stale, because it updates when someone remembers.
It breaks on provenance. A cell says €6,750,000. Which statement did that come from? Usually the answer lives in the head of whoever typed it. When that person is on holiday and a number looks wrong, there is no route back to the source, only a folder of PDFs and a filename convention.
It breaks on restatements. An administrator revises a quarter. In a workbook you overwrite the cell. The prior figure is gone, and so is any record that it changed, so a number you reported to the family last quarter can no longer be reproduced.
It breaks on the bookkeeping seam. The positions live in the workbook and the books for each legal entity live with the Steuerberater, in a different system, on a different cadence. Reconciling them at year end is archaeology, and it is the single most reliably painful week of the year.
And it breaks on the bus factor. One person understands both the workbook and the folder. Everything works until they leave.
What is actually hard to replace
The reason the workbook survives is that the alternatives usually fail on the same three things.
Instrument coverage. Plenty of tools model fund LP commitments well. Rather fewer also handle a SAFE, a convertible note, an option grant with a vesting schedule, a secondary purchase, and a direct equity stake, which is what a family office that has been active for a decade actually holds. A tool that covers 70% of your positions means you keep the workbook for the other 30%, and now you maintain two things.
Getting the data in. Every system is fine once populated. The work is the quarterly stream of statements from a dozen administrators in a dozen layouts, and any tool that expects clean input has moved the problem rather than solved it.
The books. Positions and bookkeeping are treated as separate categories by almost every vendor, so they end up in separate products and you own the reconciliation.
What we built instead
We built Wealth Management by Zahlenwerk around those three constraints, because they are the ones that decide whether the workbook actually gets retired.
Positions cover fund LP and SPV commitments, direct equity, SAFEs, convertibles, options with vesting, and secondaries, with multi-currency FX so the consolidated view lands in your reporting currency. Statements arrive by forwarding them to a private address, and get parsed into structured records, with every figure linked back to the page it was read from and its source type. Anything the model is not confident about waits in a review queue rather than entering the portfolio as provisional data. And the bookkeeping sits in the same system: a set of books per legal entity, with chart of accounts, bank transactions, coding, loans and interest, and an exportable accounting package.
The assistant is the part people notice in a demo, but it is downstream of all of the above. It can only answer usefully because the records underneath are structured and traceable.
What it does not fix
It does not remove the need for someone who understands the portfolio. It parses statements well and still gets some wrong, which is why the review queue exists and why it is deliberately conservative. And it will not reconcile a decade of history for you on day one.
The realistic first step is narrower than a migration: forward one quarter of statements, and compare what comes out against the workbook you already trust. If it does not match, you have learned something either way. That is roughly how every firm currently in the beta started, and it is a better test than any demo we could give you.