AI Compliance in Real Estate

The Compliance Guard: Fair Housing Scanning of What Ace Writes

Ace scans the messages it drafts — emails, texts, call scripts and market updates — against a library of prohibited Fair Housing and licensing-boundary patterns. Where the scan runs depends on the path: email Ace drafts inside Follow Up Boss is rewritten and rescanned before it reaches you, Automation and Autopilot sends pass a code-level gate before they go out, and the Fair Housing directive is part of every Ace chat and voice response. The scan catches obvious violations (racial or ethnic references) and subtle steering language that agents often use without realizing the legal risk.

Steering Language: What Gets Flagged

The Compliance Guard specifically targets steering language — phrases that direct buyers toward or away from neighborhoods based on protected class characteristics. The patterns include 'good schools' (implies family status preference), 'up-and-coming neighborhood' (can imply racial transition), 'family-friendly area' (familial status), 'safe neighborhood' (can correlate with racial demographics), 'quiet community' (can imply age or family composition preference), 'diverse area,' 'ethnic restaurants nearby,' and 'church community.' What happens next depends on the path: a templated Automation email that fails the scan is blocked, with a note in Follow Up Boss and an alert to your admin; email Ace drafts in the embed is rewritten and rescanned before you see it.

AI Disclosure Requirements

Multiple states are implementing requirements for disclosure when AI assists in drafting communications to consumers. Ace ships disclosure templates for email and SMS, and stamps them by path: Autopilot sends and templated Automation emails carry the AI disclosure at send time, always. Drafts an agent reviews and sends themselves carry none by default, because once an agent reviews a message the agent is the author of record; an account can opt in to the footer on those too. The AI-written send action inside an Automation is scanned for compliance but is not disclosure-stamped today, and voice has no automatic spoken disclosure. Check your state's rules and your broker's policy for anything beyond that.

Protected Classes and Discriminatory Language

Beyond steering, the Compliance Guard monitors for language referencing the seven federally protected classes under the Fair Housing Act: race, color, religion, sex, national origin, familial status, and disability. Ace is also instructed to describe properties and neighborhoods in terms of physical features, amenities, proximity to services, and market data rather than demographic characteristics of current residents.

Compliance on Automated Sends

Automations and Autopilot can run without an agent in the loop, so the scan is a hard gate in code there. A templated Automation email that fails is blocked; an AI-written Automation send is rewritten before it goes out, with generic wording if the rewrite also trips the scan; and Autopilot withholds a send that trips a high-severity flag. An SMS Automation never sends a text on its own — it produces a compliance-checked draft plus a task for the agent to review and send. On the Claude / ChatGPT connector your AI client writes the reply, so Ace asks it to run the scan before you send; the agent owns that send.

Frequently Asked Questions