your brokerage's AI, governed.
Your agents are already using AI to write listings, ads, and newsletters. That AI is fast — but the moment it reaches a client, your brokerage owns the risk. This is a plain-language walkthrough of the system JNOW puts in place so your AI stays fast and safe: it checks every message before a client sees it, stops anything that breaks the rules, keeps a permanent record, and hands you proof you can show a regulator or your insurer.
What it does, in one breath
Think of it as a safety check on your AI. Every time your AI writes something that will reach a client, the system reads it first. If it breaks a rule, the system stops it. If it's fine, it goes out — and either way, the system writes down what happened in a logbook nobody can erase.
It's like spell-check, but instead of catching typos it catches legal and fair-housing problems — and it keeps the receipts.
The one idea to remember: your AI can't send a client anything unless it has been checked — and there's always a record proving it was.
Why this matters to you
Two things changed the game for real-estate brokerages:
- Texas passed a new AI law — TRAIGA (the Texas Responsible Artificial Intelligence Governance Act). It took effect January 1, 2026. It sets rules for how businesses use AI.
- The old rules still apply too — TREC advertising rules and fair-housing law. AI doesn't get a pass. If an AI-written ad steers, discriminates, or misleads, that's on the brokerage.
Here's the hard part: your agents use AI on their own, but the brokerage carries the fallout — your license, your E&O insurance, your name. One bad AI-written line in a listing can become a fair-housing complaint.
The good news: the Texas law includes a "safe harbor." In plain terms, if you can show you had real controls in place and kept records, you're in a far stronger position if anyone ever raises a question. This system is built to earn you that safe harbor — and to hand you the paperwork that proves it.
Think of it like a seatbelt and a dashcam. The seatbelt prevents the harm; the dashcam proves you were driving responsibly.
How it works — five simple steps
Every AI message meant for a client passes through the same five steps. You don't run these — they happen automatically, in under a second, every time.
- A checker reads it first. Before anything reaches a client, a "screener" reads the AI's draft and checks it against the real Texas rules — the ones we keep in a rulebook that stays up to date.
- A gate blocks the clearly-illegal stuff. Some things are never allowed (for example, content that encourages harm). A gate catches those before the AI even runs and stops them cold.
- Nothing goes out unless it's been checked. The AI has to attach a little stamp that says "this was screened." If the stamp is missing, the system holds the message back instead of sending it. Like a kitchen where no plate leaves without the chef's sign-off.
- Everything is written in a logbook you can't edit. Every check, block, and approval is recorded in a permanent, add-only log. Nobody — not even us — can go back and change it. That's what makes it trustworthy as evidence.
- A report turns the logbook into proof. Once a month, the log becomes a clean, dated report that shows exactly what was screened, what was blocked, and what was held — the document you can hand to a regulator or your insurer.
How you use it, day to day
The best part: you barely have to do anything different. The safety is built in underneath.
| You want to… | What you do |
|---|---|
| Write a listing, ad, or newsletter with AI | Use the AI like normal. The checking happens automatically before anything reaches a client. |
| Prove you're compliant | Open your monthly TRAIGA-Readiness Report and read it — or hand it over. It's already written. |
| Answer a regulator or your E&O insurer | Give them the report. It maps each Texas rule to the control that covers it, with the real numbers. |
| Keep up with the law | Nothing — we watch the Texas rules for changes and tell you when something new applies to you. |
In short: the agents keep working the way they like, and the brokerage gets a shield and a paper trail without adding a job for anyone.
The report: your proof
Once a month you get a TRAIGA-Readiness Report. It's short, dated, and written to be handed to someone official. It says, in one place:
- Each Texas rule that applies to you — and the exact control that covers it.
- The real numbers — how many AI messages were screened, how many were blocked, how many were held back for a person to look at.
- What's still a work in progress — stated honestly, not hidden. (More on that next.)
Because it comes straight from that un-editable logbook, it isn't marketing — it's evidence. That's the whole point of the safe harbor.
What it does not do (the honest part)
We'd rather tell you the limits up front than let you over-trust it.
It proves the checking happened — it is not a lawyer. The system makes sure every client-facing message got screened and leaves proof. It does not guarantee that every judgment call was legally perfect. Hard calls still go to a human.
- It cannot promise a piece of content is 100% legally correct — only that it was screened against the rules and recorded.
- Some things still need human review. The system flags them and holds them; it doesn't pretend to decide them.
- It is a governance record, not legal advice. For a real legal question, you still call your attorney.
Being honest about this is a feature, not a weakness. A document that over-promises is worse than useless if anyone ever checks it — honesty is exactly what makes your proof hold up.
The whole thing in one picture
Here's the same five steps, laid out as a single flow:
Your agent asks the AI for a listing ↓ Gate — is this clearly against the rules? → if yes, BLOCKED (and logged) ↓ if no The AI writes the draft → the screener checks it → attaches the "screened" stamp ↓ Gate — is the stamp there? → if missing, HELD for review ↓ if yes The client gets the message — and every step above is written in the logbook ↓ Once a month, the logbook becomes your report — your proof
The short version
If you remember five things:
- Your AI can't reach a client without being checked first.
- Clearly-illegal requests get blocked before they run.
- Every check is written in a logbook nobody can edit.
- Once a month you get a report that proves it — ready to hand to a regulator or insurer.
- It's honest about its limits: it proves the checking happened; it isn't a lawyer.
That's the deal: your agents keep their speed, and your brokerage keeps its shield.
want the technical breakdown — the register, the gates, the event log, how the pieces connect? ask us for the technical reference.
See what a governed AI system looks like for your brokerage.
Book a short call. We'll walk through the register, the gates, and the report — and what it would take to run this against your actual listings and ads.
book a call →