CRE Market Research

What it extracts

16 extraction fields

Document Metadata

The master record - one row per uploaded file with title, publisher, type, date, period, geographic scope, product types, named submarkets, thesis, and author - the join anchor every other extraction row ties back to.

Markets and Property Types Covered

The coverage index - one row per market and property-type combination the document materially addresses, with coverage depth and a configurable priority-market flag - so a reader can filter the corpus by what a document actually covers without opening it.

Key Statistics

The main fact table - every quantitative figure with its unit, market, submarket, property type, period, and citation, broken out exhaustively (direct versus sublease vacancy, asking versus effective rent, quarterly versus YTD absorption, construction status) using categorical fields that keep the interpretive nuance queryable.

Notable Quotes

Verbatim analyst, broker, and executive statements that take a position, captured word-for-word and classified by quote type, so they can be pasted straight into a memo without re-reading the source.

Forward-Looking Views and Forecasts

Forward-looking statements and projections, each tagged with the analyst's own confidence language (Certain, Probable, Possible, Speculative) and direction, kept separate from historical statistics so past facts and future opinions never mix.

Risks Flagged

The risks, headwinds, and downside scenarios the author flags, in the document's own severity framing - the concerns an analyst weighs, captured as the author stated them rather than as the agent's own assessment.

Themes Covered

A controlled-vocabulary tag list of the major themes the document addresses (flight to quality, supply, cap rates, capital markets, and the rest) with coverage depth, so the corpus is filterable by topic.

Recommended Exhibits

An inventory of the charts, tables, and maps worth pulling into a memo or investor deck, each with two or three key takeaways so a downstream reader gets the point without opening the source.

Key Lease Transactions

Named lease deals with tenant, property, square footage, and lease type - the comparable leasing evidence, one row per transaction, distinct from aggregate leasing volume and from directional tenant signals.

Key Sales Transactions

Named sale deals with property, buyer, seller, price, price per SF or unit, and cap rate - the comparable-sales evidence that anchors pricing, one row per transaction.

Notable Tenant Activity

Directional tenant signals - moves, downsizings, expansions, renewals, sublease adds - that indicate market health even without full lease detail, each tagged positive, negative, or neutral from an owner's perspective.

Confidence and Sourcing Assessment

A per-document weighting call - source credibility, currency, data density, and any promotional or conflict bias - so a downstream reader or query layer can de-emphasize stale or self-interested sources.

External Benchmark Cross-Check

A web-searched verification layer that compares the document's headline figures (vacancy, absorption, rent, cap rate, construction, employment) against the most recent public benchmark from a named source and flags each Consistent, Divergent, Stale, or Unverifiable - so a possibly-dated report is stress-tested, not trusted.

Live Market Context Snapshot

A web-searched current read on the document's primary markets and property types - latest vacancy, rent, cap-rate, construction, and demographic or employment signals with public sources - so a reader can gauge how well the document's picture still holds today.

Portfolio Relevance and Routing

Pre-computed retrieval and routing tags - relevant strategies or funds, use cases, markets touched, suggested internal distribution, and a use-by date - configurable to your firm, so the corpus is filterable by how your team actually consumes research.

Market Intelligence Summary

A neutral one-look synthesis for fast triage and review of extraction quality - central thesis, what changed, headline numbers, implications for owners and investors, key risks, and a suggested next step - drawn from the extraction columns and the document, not a fact source itself.

Where it fits

What it needs

What it produces

Review model

A research analyst reviews the confidence-and-sourcing call, the web-searched benchmark cross-check, and the cited figures before the corpus is trusted or a summary circulates. The agent normalizes and cross-checks; it does not replace the analyst's judgment on which sources to weight.

Who uses it

Volume fit

Grounded in

• NCREIF Property Index and NCREIF/PREA Reporting Standardsverified as of 2026-07-22
• U.S. Bureau of Labor Statistics (BLS) employment seriesverified as of 2026-07-22
• U.S. Census Bureau population and demographic estimatesverified as of 2026-07-22
• Analytical stance - a market report is a point-in-time, sometimes promotional, third-party viewverified as of 2026-07-22

Changelog

• July 2026