AI Property Consultant · Personal build, in daily use · Gold Coast, AU

I manage my own rentals with an AI I built

Before Yo-Da Lai sells a workflow to a property agency, it has to survive his own rentals first. This is that system: it answers real Facebook Marketplace enquiries in his voice, tracks the rent and inspections, and never sends a thing without him. Below is a short clip of it drafting a reply to a real enquiry, then sending only once he approves.

Published July 2026 · Built and written by Yo-Da Lai · Gold Coast, Australia.

Watch it answer a real enquiry

A prospective tenant messages one of the listings: "is this still available?" The AI consultant reads the thread, drafts a reply in the same voice Yoda uses with his current tenant, and waits. He reads it, approves it, and it sends. The whole loop is unedited: the enquiry arrives, the AI drafts, a human approves, the message goes.

Real clip, about 90 seconds: the AI drafts a reply to a rental enquiry, and it only sends once a person approves.
  1. An enquiry lands on a live listing.
  2. The AI drafts a reply in Yoda's voice and asks to send.
  3. He approves — and only then does it go.
  • The enquiry: "Hello, is this still available?" — a prospective tenant, on a live Marketplace listing.
  • What the AI drafted: "Hi — yep, still available. Viewings from after the 17th if you want to book one." Written in Yoda's voice, learned from his real threads.
  • Before it sent: the system held the draft and asked him to approve, edit, or skip. He approved it. Nothing reached the tenant until he did.

Full transcript: Hey, just wanted to show you a quick example of one of the workflows I built myself to make managing my property a lot easier, so it doesn't take so much time out of my day. A common issue I keep facing is that so many people message just to ask if a place is still available — when the information is already on the post — and you don't want to spend your day answering "yep, it's available" only for them to ghost you or ask more questions. So here's the system I built. You can see it's asking me for approval before it sends the response, and I worked through a couple of iterations to improve the wording until it read the way I like. It's based on how I actually talk to my current tenant: it reads through the real conversation I've had with them, learns from it, and drafts the replies. I get to improve the draft, and once I'm happy with it, I tell it to send. Approved — and the message is sent.

What was I actually solving?

Every listing draws the same trickle of "is this still available?" messages, most of them asking what the ad already says. Answering by hand, or leaving people on read until he got to his phone, was quietly losing him tenants — so he built the fix for his own rentals first, and only offered agencies what had already survived his own doors.

What does it do, day to day?

Answering enquiries is one job of several. On Yoda's own portfolio the same system tracks the rent and bills as they fall due, plans inspections, checks lease renewals against the market, and helps when a place goes on the market — drafting the listing copy and the replies, never posting or sending on its own. It learns how he actually communicates by reading his real threads with his current tenant, so a drafted reply sounds like him, not like a bot. Everything it produces is a draft until he approves it.

  • Tracks rent and bills as they fall due, so nothing slips past its date.
  • Plans inspections and flags lease renewals, checked against the market.
  • Drafts replies to rental enquiries in Yoda's own voice — he approves, then it sends.
  • Helps list a property: drafts the ad copy and the replies, never posts on its own.
  • Learns from his real tenant threads, so every draft sounds like him.

What changed once it was running?

The trickle of enquiries stopped being an interruption. A message lands, a reply is drafted in Yoda's voice within moments, and all that's left is a glance and a tap. The rent, the bills, the inspections and renewals are tracked as they come due instead of living in his head and a spreadsheet. He still makes every real decision; he has just stopped doing the typing and the remembering. It's the same shift he sells to agencies, except here he is the one who felt it first.

BeforeAfter
Enquiry replies: typed by hand, whenever he next picked up his phoneDrafted the moment a message lands; he approves and it sends
The reply's voice: only sounded like him if he wrote itLearned from his real threads — every draft sounds like him
Rent, bills, inspections: tracked in his head and a spreadsheetTracked as they fall due, surfaced before they slip
Who decides: him, doing all of itHim approving — the system does the legwork

What happens before anything sends?

Nothing the consultant writes reaches a tenant, an owner, or a listing without Yoda. Every reply, every post, every update is a draft first: the system shows him what it wants to do and waits for a yes. It has no access to move money and doesn't touch anything to do with trust accounting — it drafts and organises, and a person approves anything that matters. When something falls outside its rules, it stops and hands the decision back instead of guessing, and every action it takes is logged. It's the same human-approval pattern behind every workflow he installs for an agency — proven here, on his own properties, before it's offered to anyone else.

Straight answers

Is this used by a real agency, or just Yoda's own rentals?

Right now, his own. This is his portfolio's system and his proving ground — he runs every workflow here before offering it to an agency. The agency version is the same engine, built around the software a team already uses.

Does it touch trust accounting or move any money?

No. It has no access to move money and nothing to do with trust accounting. It drafts replies, tracks dates, and organises information; a person approves anything that matters.

Who approves what the AI writes?

On Yoda's own portfolio, Yoda. For an agency build, the property manager or principal. Nothing sends, posts, or triggers until a person signs off.

Does it ever send a message on its own?

By default, no — every draft waits for approval. Only once he has reviewed how it replies can he let it run a specific, low-risk reply on its own, and even then it stays inside strict rules and logs everything.

More from Yo-Da Lai

Case studies: Loantec · Youknow Clothing · Saint Clare · Batchit · Jimmi Jackson · AI Property Consultant. Back to the full portfolio · ClawConnect pricing and live deployments. Book a free build review at yodalai.xyz or email yoda@yodalai.xyz.