Two agents, one Sunday night
Carla and Dev collected the same Sunday leads. She spends the night on inbox triage. His AI agents already replied. Same market, same skills, same Sunday.

Carla and Dev work at different brokerages in the same metro. Both held open houses this Sunday for comparable listings. Both collected about thirty sign-ins, and both went home to the same pile of digital exhaust: fourteen portal inquiries from the weekend, a text thread with a lender, two sellers wanting updates, and a buyer client who found four new listings and wants opinions on all of them.
It is 9 pm on Sunday. This is where their weeks diverge.
Carla's Sunday night
Carla pours a glass of wine and opens her laptop, because Sunday night is when she does what she privately calls inbox triage and what a labor economist would call a second job. She copies the open house sign-ins from the paper sheet into her CRM, or means to; she gets through eleven names before deciding the rest can wait. She answers the four portal inquiries that look serious, using variations of the same paragraph she has typed a thousand times. She texts one seller an update and decides the other can wait for Monday. The buyer with four listings gets a "great finds! let's discuss tomorrow" that both of them know means Tuesday.
By industry standards Carla is diligent. The research on lead response has been depressing and stable for fifteen years: most inquiries never receive any response at all, and the average response time for those that do is measured in hours while the winning window is measured in minutes. Carla responds to most things within a day. That puts her in the top quartile of her profession and still guarantees she loses leads every single week, silently, to whoever answered first. She will never know which ones. That is the cruelest part. Lost leads do not send a notification.
“She will never know which ones. That is the cruelest part. Lost leads do not send a notification.”
Dev's Sunday night
Dev watches television.
This is not because Dev cares less. It is because Dev's brokerage, like a growing number, has restructured the after-hours work around a small crew of AI agents that operate inside the team's shared workspace, and Sunday night is exactly the load they were brought in to carry.
While Dev watches television, an agent is working through the weekend's inquiries. Not with an autoresponder blast; the industry tried that a decade ago and buyers learned to smell it instantly. The agent reads each inquiry, pulls the property details, checks the buyer's questions against the listing data, and drafts an individual reply: yes, the HOA allows dogs under forty pounds, here is the fee schedule, and Dev has openings Tuesday at 4 or Wednesday at 5:30 for a showing. Routine replies go out within minutes of the inquiry arriving, at 9 pm, because a buyer scrolling listings on Sunday night is a buyer at peak intent. Anything involving price, negotiation, or a question the agent isn't confident about gets held in a queue with a drafted response for Dev to approve, and Dev's phone pings only for the one inquiry that mentioned being a relocation buyer with a three-week timeline, because that one is worth interrupting television for.
The open house sign-ins went into the system as they were captured, and a second agent has already cross-referenced them against the brokerage's existing contacts. Two of Sunday's visitors inquired about a different listing eight months ago. That context is now attached to their records, and the follow-up drafts reflect it. Nobody typed anything into a CRM. On this team the CRM stopped being a chore you owe the system and became a thing the system simply knows.
The two sellers each got their weekly update drafted from actual data: showings this week, feedback collected from the showing agents (an agent pinged each one Monday through Saturday and actually got answers, because it never forgets to ask twice), days on market against comparables, portal view counts. Dev reads both drafts Monday at 8 am, softens a sentence in one of them because he knows that seller is anxious, and sends. Reviewing takes four minutes. Writing them used to take forty.
The buyer with four saved listings got a reply Sunday night too, with the agent's honest first pass: two of the four are already under contract (it checked), one backs onto the arterial road the buyer previously rejected a house for, and the fourth looks genuinely promising and is open Thursday. Dev wakes up to a proposed Thursday showing already accepted.
The part people get wrong
The reflex objection is that real estate is a relationship business and this mechanizes the relationship. The teams doing it well have concluded exactly the opposite, and the logic is worth sitting with.
What buyers and sellers experience as relationship is mostly responsiveness plus judgment. The responsiveness half was always being rationed, because a human agent has one phone, one calendar, and a family. Every hour Carla spends copying sign-ins is an hour of rationing. Dev rations nothing. Every inquiry gets a real answer fast, every seller gets a substantive update, every showing gets feedback chased down, and the humans in the operation spend their hours on the half that actually is the relationship: the pricing conversation, the negotiation, the talk-down when an inspection report scares a first-time buyer, the walk-through where thirty years of instinct notices the water stain.
Dev's team also runs a rule that tells you everything about where this is going: any message that could change what a client decides gets a human's eyes before it leaves the building. The agents draft, verify, schedule, chase, and reconcile. People persuade and advise. The line is drawn in judgment, not in hours.
Carla will hear about all this eventually, probably at a conference, probably framed as a threat. It isn't one, or at least not the one she will expect. Nobody is coming for the part of her job she loves.
Monday morning, Carla has nineteen sign-ins still on paper. Dev has a showing booked for Thursday. Same market, same skills, same Sunday.
How many Sundays does that gap take to become a career?
FAQ
Small hops. Big leap.
Every drafted follow-up, every synced table, every brief that writes itself is one small hop. Together they change how the team moves. Early access is open.
Get started
Alex Shershebnev
Alex Shershebnev is a seasoned AI engineer and technology leader with over a decade of experience in AI, DevOps and MLOps. He is currently Lead DevRel at Zencoder, an AI coding assistant, and one of the founding members of the company, where he has spent the last two years shaping both the product and its developer ecosystem. Alex has spoken at more than 50 international conferences, establishing himself as a recognized voice on AI for coding, secure and responsible use of AI in software development, and the future of developer workflows.
Author
Alex Shershebnev