Services / Obligations Library
Every requirement, structured.
Twenty-five thousand discrete obligations extracted from the six guides and the federal mortgage regulations — filterable, sourced, and linked back to the text.
Why this matters.
A compliance officer who needs to know what the Fannie Mae Selling Guide requires of a lender at closing has two options: read the whole guide until they find it, or ask someone who already read it. Neither option scales. The guide is over a thousand pages, and the someone-who-already-read-it is also their colleague who is also too busy.
The obligations library is the structured index that should have existed alongside the guides all along — every discrete “must,” “shall,” and “is required to” extracted, tagged, and linked to the section it came from. It does not replace the primary text. It gets you to the right section in seconds instead of minutes.
What this looks like.
How we do it.
Section-level extraction
Each guide and regulation is processed section by section. The AI reads the section text and identifies every discrete obligation — a “must,” a “shall,” a “required to” — and records the exact source anchor.
Tagging and party attribution
Each obligation is tagged by topic (from the same closed taxonomy used across the platform) and attributed to the responsible party: lender, servicer, or broker. Topic and source drive the filters; the party is shown on every obligation.
Staleness-driven refresh
When a guide is updated, the affected sections are re-extracted and existing obligations are updated or replaced. The library stays current as the source guides change.
What's in the product today.
Extracted from six GSE and agency guides — Fannie Mae, Freddie Mac, VA, FHA, USDA Rural Development, Ginnie Mae — plus federal mortgage regulations including Reg Z, Reg X, Reg B, HMDA/Reg C, SAFE Act, and BSA
Filter by source guide or regulation and by topic tag; every obligation is attributed to the responsible party — lender, servicer, or broker
Each obligation links back to the specific section it came from, so you can read the primary source in one click
Confidence rating on every extracted obligation so you know how firmly it is grounded in the text
Roughly 25,000 obligations across the full corpus, browsable without a keyword in mind
Framed as a shortcut to the source — obligations are an index, not a substitute for the primary text