How this gets made

How research and drafting work

This is an AI-assisted publication and we would rather say so than be found out. Automated systems gather vendor pages and filings, extract structured data, assemble first drafts against a research brief, and check finished pages against our house rules at a scale an editorial desk could not sustain. They do not decide which claims are strong enough to publish, which sources are credible, or which page is ready. Those are editorial judgments and the desk makes them. Nothing publishes unread.

What the software refuses to publish

Our standards are written as tests that run before a page can ship. The build fails if a page is missing a published date, a modified date, a description or a canonical URL; if a decision page lacks a Quick Answer or its Quick Answer runs past eighty words; if a figure in the text has no entry in the Sources list; if an internal link is broken; or if the structured data does not match the visible table.

Where the facts come from

• Vendor pricing and product pages, read on the date shown in the table, linked from the cell.
• SEC filings: Form D and Form ADV, read directly rather than through an aggregator.
• Fund documents and ILPA templates.
• Named practitioners speaking on the record.

Verification cadence

Every table row carries a "Last verified" date. Rows are re-checked at least quarterly and whenever a vendor tells us something changed. A vendor can request a re-check at any time.

When we get one wrong

We will publish errors. A publication at this volume that claims otherwise is either not looking or not telling you. Corrections are made in place with a dated note, and substantive ones are listed on the corrections page.