Trust & Safety
How SRRF's AI works β and what it will never do.
Real estate runs on trust. Here's exactly how your AI Concierge operates, the guardrails that keep it honest, and the controls that keep you in charge.
The AI behind your Concierge
Your AI Concierge is powered by Anthropic Claude β one of the safest, most rigorously tested large language models on the market. This is the one place we disclose the model, because trust matters more than keeping it a secret.
It's not a general-purpose chatbot. We've built it specifically for real estate: your voice, your knowledge base, your listings, your rules β and the compliance guardrails below, which it cannot override.
It won't invent listings
Your AI only speaks to what it actually knows β the property details, prices, and neighborhood facts your realtor provided, or data from a connected source. If a lead asks about something it has no data on, it says so and hands off to your realtor. It never fabricates a listing, a comp, an amenity, or a price.
It also follows Fair Housing rules by design: it won't make qualitative claims about schools, neighborhoods, or communities. This is the single most important safeguard we ship, and it's enforced on every message.
Opt-out and HELP, honored instantly
Every lead can stop the conversation at any time. A STOP-style message β in English, Portuguese, or Spanish β permanently opts them out: the AI stops replying, pending follow-ups are cancelled, and the opt-out is time-stamped for your records. This meets TCPA and the FCC's 2025 "any reasonable means" standard.
A lead who sends HELP or INFO gets an immediate reply identifying the business, how to reach a human directly, and how to opt out β no AI generation, just the facts they're owed.
What SRRF will never do
- Never quote a price your realtor didn't provide.
- Never invent listings, comparables, or amenities.
- Never negotiate, offer discounts, or make a financial commitment β only your realtor can close a deal.
- Never sign a contract or move money on your behalf.
- Never promise timelines beyond a scheduled appointment.
- Never ignore a lead's opt-out.
You hold the kill switch
You're always in control. Pause the AI for a single conversation from your inbox, or pause it account-wide from AI Control β both take effect in seconds, and the AI resumes only when you say so. Leads keep getting captured either way; you just decide when the AI speaks.
