Automation Teardown #4, real estate edition: a buyer inquiry comes in at 9:40pm on a Saturday, and by the time anyone answers it Monday morning, they've already toured with someone else.
The manual reality
Inquiries arrive from everywhere: portal leads, the website form, texts to the number on the sign rider, replies to old listing emails. Someone answers the same questions over and over. Is it still available? What's the HOA fee? Pets? Can I see it this week? Then they guess who's serious and paste the whole thread into the CRM so an agent can pick it up. Speed matters more here than almost anywhere else, and this is the step where the time goes.
What the agent does
- One queue. Every inquiry, from any channel, lands in the same place with the source attached.
- Answers from the listing only. The agent answers factual questions such as status, fees, square footage, and showing windows, using the listing data and nothing else. That data is pulled fresh from the brokerage's feed, so when a listing goes pending, the answers change with it. If the answer isn't in the listing, it says so and hands off. It doesn't make one up.
- Qualifies with the brokerage's questions. It asks the questions the team already asks: timeline, pre-approval status, and whether the buyer is already working with an agent. The brokerage writes those questions, not the model.
- Routes with the paperwork pulled. Serious leads go to the right agent, matched by listing, area, or rotation. They arrive with the listing sheet, the disclosure packet, and open showing slots attached, so the agent's first message can be "Thursday at 5 works." Everything else gets a polite, spaced-out follow-up cadence instead of silence.
What stays human
Anything about price, offers, or negotiation goes straight to a licensed agent. The assistant doesn't interpret a disclosure, give advice about the property, or speculate about what a seller would accept. Agents approve every showing before it's confirmed. And all of it is subject to the brokerage's own compliance rules. The agent sticks to facts and logistics, the same as a good assistant at the front desk would.
The guardrails that make it trustworthy
- Every reply is traceable. Each answer is logged next to the listing data it came from, so a broker can see exactly why the agent said what it said.
- Same facts for everyone. The agent gives every inquirer the same information and never comments on who a neighborhood is "right for." Fair housing rules apply to software too.
- One switch to pause it. Brokers can turn it off at any time, and inquiries fall back to the team's normal process.
How we'd know it worked
Two numbers, measured before and after in the first month: time to first response, and the share of inquiries that turn into showings. If neither moves, the agent isn't earning its place.
When it isn't worth it
If you get a handful of inquiries a week, a good auto-reply and a fast phone answer will do the job. This pays off at volume, when the team can't answer everyone quickly by hand.
Honest sizing
This usually takes first response from hours to minutes and gives the team back a few hours a week of copy-paste and repeat answers. It's typically live in a week or two, depending on how many inquiry sources need to be connected. A brokerage with one main lead source will be on the short end of that.
We run free 20-minute teardowns of workflows like this. Start at /discuss or write to hello@fullspecstudio.com.