Family offices are using AI to eliminate the manual document work that keeps their balance sheet perpetually out of date. MyFO does this natively — and across every asset class, not just fund positions. Its AI document ingestion reads K-1s, capital call notices, distribution notices, statements, loan documents, and appraisals, extracts the underlying data, and writes it into the platform against the correct entity and asset. No re-keying, no parallel spreadsheet.

‍

The distinction matters because most tools solve one slice of this. Some are built for public markets. Some are built for LP fund positions. Some only for real estate. A few cover a bit of two. Very few can tell you what the apartment building, the shareholder loan, the stake in the family's operating company, and the art collection are worth today — let alone hold them in one view alongside everything else.

‍

‍

What is the actual data problem?

‍

It is not that any one asset class lacks a tool. It is that the tools are siloed. There are capable aggregators built purely for public equities, others built purely for LP positions, others built purely for real estate — each competent inside its own lane, none of them holding the whole balance sheet. Run three of them and you have three partial pictures and a spreadsheet trying to add them up.

‍

Take LP positions, the silo that gets the most attention right now. A family office holding fifty to a hundred private positions receives a steady flow of PDFs: capital calls with a due date and a wire amount, distribution notices with a return-of-capital split, quarterly statements with an updated NAV, and K-1s months after year end. Each one has to be opened, read, keyed into a tracker, and reconciled against the commitment. AI handles that well, and several platforms now do.

‍

But LP positions are the easy private assets. They at least arrive with documents on a schedule. The harder question is everything else on the balance sheet:

‍

• Real estate — properties held in dedicated entities, each with its own mortgage, rent roll, insurance renewal, and tax bill
• Loans and private credit — shareholder loans, notes receivable, intra-family lending, and debt against assets, all with their own amortization and maturity dates
• Directly held operating companies — a medical practice, a manufacturing business, a family holding, valued off financial statements rather than a NAV letter
• Heirlooms and passion assets — art, watches, collections, valued off appraisals that arrive every few years

‍

None of this comes through a custodian feed. Most of it arrives as a PDF, a spreadsheet from an accountant, or an email from a property manager. This is precisely the material that most reporting tools were never built to hold — so it ends up in a spreadsheet beside the platform, which means the consolidated view is never actually consolidated.

‍

MyFO ingests all of it into one entity model, and calculates gains, valuations, and LP multiples — DPI, RVPI, and TVPI — so a direct real estate holding and a venture fund sit in the same view on the same basis.

‍

‍

What does MyFO's AI actually do with those documents?

‍

• Reads and classifies the document — capital call, distribution, statement, K-1, agreement
• Extracts the fields that matter — amounts, dates, NAV, commitment, called and distributed capital, appraised values, loan balances and maturities, and cost basis and tax lot detail where the source provides it
• Links it to structure — the document and its data attach to the right entity and the right asset automatically, because folders are pre-created and entity-linked
• Surfaces it as a review step — extracted values are presented for confirmation rather than posted silently, so a human stays in the loop
• Feeds downstream — updated NAV and called capital flow into consolidated reporting, and known capital calls flow into cash flow forecasting

‍

The same pipeline is what makes onboarding fast. A backlog of statements and agreements becomes a structured portfolio in weeks rather than months of manual entry.

‍

‍

Why does structure matter more than the AI itself?

‍

Extraction is the easy half. The reason a document pipeline is useful in one platform and useless in another is where the output lands.

‍

MyFO creates the entity, its ownership relationships, and its folder structure first. Every extracted value therefore arrives with context — which trust owns it, which fund it belongs to, which reporting period it affects. Unstructured extraction into a general file store produces searchable PDFs and nothing else.

‍

You can see the difference in practice in the real estate example above: the documents and the numbers are the same records, so a loan maturity buried in a PDF becomes a deadline in the forecast.

‍

‍

How does MyFO compare to Canoe Intelligence and Arch?

‍

Canoe and Arch are both good at what they do, and MyFO partners with both where a client needs them.

‍

The difference is scope. Canoe and Arch handle alternative asset document ingestion and then hand structured data onward — you still need a reporting and entity layer to consume it, which means two vendors, two contracts, and an integration between them. MyFO does the ingestion and the reporting in one platform, against one entity model.

‍

When it makes sense to combine them:

‍

• Canoe — for families that specifically want to use Canoe's own LLM and document processing
• Arch — for families that want a service layer, with a team handling the document workflow on their behalf

‍

In both cases MyFO acts as the top layer: partners handle the private markets feed, MyFO holds the entity structure, reporting, forecasting, and workflows. MyFO also pulls from Juniper Square for LP and partnership management, which is the fund administration system most MyFO clients already use.

‍

‍

FAQ

‍

Does AI post data without review? No. Extracted values are surfaced for confirmation, with a correction step before anything is committed.

‍

Which documents can MyFO ingest? Capital call and distribution notices, K-1s, investment and custodial statements, subscription and partnership agreements, loan and mortgage documents, appraisals, property and operating-company financials, and CSV or Excel exports.

‍

Can AI handle assets that aren't fund positions? Yes, and that is the harder half of the problem. Real estate, shareholder loans, directly held operating companies, and collectibles rarely arrive through a feed — they arrive as appraisals, statements from an accountant, or emails from a property manager. MyFO ingests those into the same entity model as LP positions rather than leaving them in a side spreadsheet.

‍

Do we need a letter of authorization to get data in? No. Statement and custodial-export upload is the default path, with aggregators and institutional feeds layered in where available.

‍

Does MyFO replace Canoe or Arch? It can, for offices that want one platform. It also integrates with both when a family prefers Canoe's LLM or Arch's service model.

‍

‍

Want to see how this works for your family office? Book a call and we'll show you exactly what it looks like at your scale.