Somebody asked me this on a call the other day, and I could tell he was hoping for a quick yes. He'd just watched a demo where an AI voice agent dialed a list, sounded like a real person, and booked an appointment. His question was simple: "Can I just turn that loose on my seller leads?"
And my answer was the honest one, which is: it depends, and the part it depends on is the part the demo never shows you.
I want to walk through how I think about this, because I build these systems for a living and I still slow down every single time before I let a machine touch a phone. Before I go any further, let me say the thing plainly: I am not an attorney, nothing here is legal advice, and the rules change and differ by state. What I can give you is the operator's version — the questions I make myself answer first, so I don't build something that gets me in trouble.
The one word the whole thing hangs on: consent
Different situations get treated differently, and I'll acknowledge that up front — some calls are fine, some are a problem, and the line between them is not about how human the voice sounds. It's about consent.
Here's the way I understand it, in plain English. There's a federal law — folks call it the TCPA — that governs automated and pre-recorded calls to people's cell phones. And the general shape of it, as I understand it, is this: if you're using an automated system or an artificial or pre-recorded voice to call somebody's cell, you're generally supposed to have their permission first. From what I understand, an AI voice counts as that kind of voice. It doesn't matter that it sounds like your cousin. What seems to matter is that it's generated, not a live human dialing. But don't take my word for the legal line — that's exactly the kind of thing to run past an attorney.
That's the piece the demo skips. The bot works. The bot sounds great. But the question isn't "does it work," it's "did this person say I could call them like this." A cold seller who's never heard of you, whose number you pulled off a list, has not given you that permission. A lead who filled out your form and checked a box saying "yes, text and call me" is a very different situation.
And the reason I take it seriously is the same reason I take the numbers seriously in everything I do: the penalties people quote for getting this wrong aren't small. The figure I've seen most is around five hundred dollars a call, and more than that if it's treated as willful, with no cap on how many calls add up. Now, I can't promise that's the number that would apply to you — the exact amount is an attorney's call, not mine — but you don't have to do much arithmetic on a ten-thousand-record list to see how a bad campaign turns into a figure that ends a business. That's enough for me to know it's not a corner I cut.
Where I actually let AI call, and where I don't
So does that mean AI calling is useless for real estate? Not at all. It means I put the machine where consent already lives, and I keep it off the cold list. Let me show you the line I draw, because it's simpler than it sounds.
I let AI handle the warm calls. Somebody opts in on my site. Somebody replies to a text and asks me to call. Somebody I've already got a relationship with and clear permission to reach. Somebody calling me — an inbound call has a completely different feel to it, because they started the conversation. That's where an AI agent earns its keep: it picks up in the first ten seconds instead of the third day, it qualifies, it books the appointment, and it never gets tired at 9 p.m. That's real value, and it's the kind of thing I've written about in how I use AI to follow up with seller leads without dropping the human touch.
The cold list, I keep human. A real person, dialing a number they have a legitimate reason to call, having a real conversation. I don't hand a raw, never-heard-of-me list to a bot and let it dial away. My thinking is this: the cold outreach is exactly where the consent question is the shakiest, so that's exactly where I want a human making judgment calls, not a machine making the same mistake ten thousand times fast.
This is really just the difference between an AI agent and plain automation showing up in a place that matters. Automation does the same thing every time, whether or not it should. A good operator decides whether it should. When the stakes are legal, you want the deciding to happen before the dialing, and you want a person in that loop.
If you want the fuller picture of how I split the work, I laid it out in what I hand off to AI agents and what I still do myself. The short version: I hand off the tireless, repetitive, permission-in-hand work, and I keep the judgment for myself. Cold calling a stranger is judgment work.
What I'd have you check before you flip the switch
Alright, here's the practical part — the checklist I run in my own head, offered as questions to bring to somebody who actually gives legal advice, not as the last word.
- Do I have documented permission to call this person this way? Not "they're on a list I bought." Actual, recorded consent — a checked box, a reply, an inbound call. If I can't point to it, the AI doesn't dial.
- Is this a cell phone? Because the rules for automated and pre-recorded calls to cell phones are the ones I'm most careful about, and most seller numbers these days are cells.
- Am I honest that it's AI? My instinct, and I think the decent thing regardless of the law, is to not pretend a bot is a person. Disclosure is both cleaner and, from what I understand, often expected.
- Have I scrubbed against the do-not-call rules and stuck to reasonable calling hours? These are basic, and they apply to humans too.
- Have I asked a real professional about my state? This is the big one. The federal shape is one thing; your state may layer more on top. Knowing the rules where you operate is just part of doing this the right way — the same reason my friend Chris Albin up in Illinois wrote about whether wholesaling is even legal in his state. The law is local, and it's worth the phone call to find out.
I run that list every time, and honestly it takes about five minutes. Five minutes against a number that could end a business is the easiest trade I make all week.
Why I keep the machine on a leash
Here's where I land, and it's the same place I land on most of this AI work. The technology is genuinely good now. An AI agent can hold a warm conversation, book the appointment, and free you up to go do the deals. I'm not anti-machine — I build the machines.
But my people are the folks who have to struggle to put it together, and the last thing I want is for one of you to watch a slick demo, point a bot at a cold list, and wake up to a problem that a five-minute check would have caught. Ultimately, the tool isn't the risk. Turning it loose without asking who gave you permission is the risk.
So point AI at the warm calls, where consent already lives, and it'll pay you back every day. Keep it off the cold ones until you've talked to somebody who knows your state. That's the goal — real work getting done, without the letter from a lawyer.
I don't know if that fully answers it, because your situation has details mine doesn't. But it's the honest place I'd want a friend to start.
This post is for informational and educational purposes only and is not financial, legal, or investment advice. Telemarketing and automated-calling laws (including the TCPA) are complex, change over time, and vary by state — nothing here should be relied on as a statement of what the law currently requires. Before you use any automated or AI calling system, consult a qualified attorney about your specific situation. Individual results will vary depending on your market, your resources, and your effort.
