Insights
Notes on the things we are building: data infrastructure, document extraction, MCP, and what it takes to make finance software people trust.
Live portfolio dashboards in Claude, from your own Zahlenwerk data
Claude can now build dashboards that fetch fresh data every time someone opens them. With the Zahlenwerk connector, that data can be your portfolio. How it works, what we changed to support it, and what to know before you share one.
Keep readingA small model for routine questions, and where we stop it
Most questions in a document pipeline are small. Jev answers them in about half a second, for teams that switch it on, and only decides when it is at least 0.9 sure. What it sees, where it runs, and what it is not allowed to do.
Planning for unfunded commitments in a family office
Unfunded commitment is the number a family office's liquidity plan stands on, and in most spreadsheets it is wrong. How to calculate it properly, where it goes wrong, and how to plan for the calls still to come.
A holdings register for a family office. Funds, SPVs and direct stakes in one place
Software for managing holdings is mostly built for corporate groups. A family office needs something else, a register of what it holds, through which company, on what terms, with the documents behind it.
Tracking a private equity portfolio in Excel, and when it stops working
A spreadsheet is the right answer for longer than vendors admit. Here is a structure that survives, the five things that break it, and the honest threshold for replacing it.
You missed a capital call. What actually happens next
The LPA usually gives you a cure period and charges interest. Default provisions exist but are rarely the first move. Here is the sequence, and the four failure modes that cause most misses.
Family office software compared (2026), and how to actually choose
A dozen platforms, what they cost, what each one is genuinely built for, and the four questions that decide the answer faster than any feature matrix. Including where we are the wrong choice.
IRR, TVPI and DPI, and why your three sources disagree
The metrics are simple arithmetic. The disagreement is almost never the formula. It is which cashflows each source counted, and as of when.
What an MCP server actually changes about portfolio data
Connecting a portfolio to Claude and ChatGPT is not a chat feature. It is an access-control problem, and that is the part worth getting right.
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.
Why we let the extraction model say "I am not sure"
A parser that is right 96% of the time is not 96% useful. What matters is whether it knows which 4% to flag.