An AI dialer quoted me less than one VA seat, so who's still buying VA hours?

Put an AI voice agent on a tenant whose water heater failed at 11pm and watch what happens. My VAs handle maintenance intake and the value is in the judgment call about whether it's an emergency dispatch or a Monday ticket. The $400 tool doesn't own that decision, and if it gets it wrong I pay a plumber $450 for a callout that wasn't one.

The $400 quote almost certainly excludes the list. Skip tracing and data are separate and that's where my money goes.

What I actually want is someone who does the same four things every week without me checking. CRM hygiene, sending the 30-day follow-up letters, chasing missing documents. A subscription that books appointments doesn't do any of that, and I've never seen a demo of the dull parts.

Curious about the other side of marlow's question. If the tool really is $400 and a seat is $1,500, does an agency's spread just shrink until it's selling the tool plus a human to babysit it? That sounds like a different business with a worse margin.

Reminds me of nail guns. Didn't remove framers, changed what a framer does in a day. The guy who never picked one up got slower and cheaper to replace.

Running three seats and I'll give you the number that surprised me. When we put an AI drafting tool on seller follow-up emails, output roughly doubled and my review time went up, not down, because now somebody has to catch the one message in twenty that says something we'd never say to a seller. Net it's still worth it. It did not reduce headcount.

The part of my VA's job I'd never hand to a $400 subscription is the eighth touch on a seller who said no in month one and then had a tenant trash the place. She remembers the story, calls back like a person, and knows to ask about the brother who's a co-owner. That's not qualification, that's memory.

beacon's question is the one I'm stuck on. If I build a service around this, am I selling seats or am I selling setup? Charging a client to configure their dialer, write the scripts, and then staff one human to run QA feels more defensible than renting out hours. But then I'm a consultant with a payroll problem, and I don't know if that's a better business or just a different one.