openOM

What is an offering memorandum (OM)?

An offering memorandum (OM, sometimes "offering memo" or "deal book") is the marketing document a commercial real estate (CRE) broker prepares to offer a property for sale to prospective buyers. It presents the asset, the deal terms, the tenancy, and the broker's opinion of value - typically as a designed PDF of 10-40 pages.

What an offering memorandum contains

Why OM data is an assertion, not a fact

An OM is an advertisement: a broker's opinion of value that the seller agreed to before publication. The NOI may be in-place or pro-forma; the cap rate follows from a chosen price. So every figure is an identified party's opinion as of a date - not independently verified market truth. Any system that consumes OM data honestly must record who asserted it, that it is unaltered, and as of when - and must never claim the opinion is true. Verified means provenance, not truth.

The problem: OM data is trapped in the PDF

Because the OM ships as a designed PDF, every downstream party - buyers, brokers, portals, lenders, and AI agents - re-extracts the same numbers by hand or with error-prone parsing. The work is repeated thousands of times and each copy can drift or be misread.

How openOM makes an offering memorandum machine-readable

openOM is an open (MIT) standard that embeds a machine-readable, broker-asserted, hash-verified data payload inside the OM PDF (via the same mechanism as Factur-X / PDF/A-3), and mirrors it as JSON-LD on the web. The data is extracted once at the source and consumed cheaply everywhere - with its provenance intact.

openOM vs. manual OM extraction

Manual / AI re-extractionopenOM
Where extraction happensEvery consumer, every time Once, at the source
ProvenanceLost - who asserted what is unknown Recorded - assertedBy + assertedDate
IntegrityUnverifiableHash-verified (RFC 8785 + SHA-256)
AI reliabilityHallucination-prone re-parse Deterministic read of a verified payload
Cost at scaleRepeated per documentNear-zero downstream

Next steps

openOM is published by Vervelio Labs. The engine is deterministic and inference-free; every payload is an assertion, never a fact.