Reviewing extracted data used to mean opening every document. Now it doesn’t.
Extraction is the easy part. When capital statements come in from dozens of GPs each quarter, the hours go into checking the output, and even a 90% clean rate means opening all 100 documents to find the 10 that need attention.
Our latest Mantle Portal update fixes exactly that. The new line-item extraction review separates clean capital statements from the ones that genuinely need your attention, shows you every number next to its source, and learns from your corrections so the same fix never has to be made twice.
1. Discrepancy Flagging: Review 10 documents, not 100
Not every PCAP looks the same in the queue anymore. Mantle Portal now separates capital statements that came through cleanly from those flagged for a closer look, whether that’s an ambiguous label, a value that doesn’t foot, or a discrepancy within the statement itself.
Here’s how it works in practice:
100 capital statements arrive this quarter. 90 came through without issues, and your team can move past them with confidence. The 10 that need a judgment call are surfaced automatically, and Mantle Portal tells you exactly why each one was flagged.
Your reviewers spend their time where judgment is actually required, and nowhere else.
2. Two-Table Review: Every number, one click from its source
When you open a flagged document, you now get a two-table view.
On the right: every raw line item Mantle Portal pulled from the PDF, exactly as it appears in the source, with its description, value, and how Mantle Portal categorized it. On the left: the normalized view, the clean, structured data that feeds your portfolio reporting.
This pairing matters because a single normalized category can be built from multiple source line items. Interest income/expenses in your normalized view might be the sum of two separate lines from the statement. The two-table layout shows you exactly what contributed to each number instead of asking you to trust that the math is right.
Click any value in either table and Mantle Portal highlights the corresponding text in the source PDF, and vice versa. You’re never more than one click away from verification.
Disagree with a categorization? Say a line item was filed under Other and it should be Partnership Expenses. Recategorize it directly in the review screen, or mark it as an unlinked line item if it’s not relevant to your reporting. The change applies as soon as you save.
3. Learned Categorization: Correct it once, then never again
Here’s where it compounds, and this one reflects something real about how private markets work.
Every family office books line items slightly differently. Whether fund costs get recorded under Investment Expenses or Partnership Expenses isn’t an extraction error. It’s an accounting decision specific to that office, that accounting team, and sometimes that fund. There’s no universal right answer, so your software shouldn’t pretend there is one.
Here’s how this works in practice:
Last quarter, your team recategorized Fund Cost as Partnership Expenses on a capital statement. This quarter, the same fund’s statement arrives from the same GP. Mantle Portal has already made that call for you, applying your preference automatically on the new extraction.
Preferences are stored by fund and document type, so the reconciliation work your team does once pays off every quarter after.
Why this matters for investors
Whether you’re a family office tracking dozens of fund commitments or a fund-of-funds processing hundreds of capital statements a quarter, the goal is the same: less time on operational overhead, more confidence in the data underneath your reporting. AI extraction only earns that confidence when you can verify it, correct it, and trust it to remember. That’s what this update delivers.
Want to see more? Subscribe to the Mantle blog below for product updates delivered straight to your inbox. If you’d like to see the full platform in action, book a demo.