How to use AI in real estate: seven practical jobs

AI is used in real estate for seven practical jobs: the first reply to an enquiry, listing copy, listing photos and video, client records, matching listings to buyers, access to land registry data and visibility in search. In each of them AI prepares a draft or pulls scattered information into one place. The agent still sets the price, runs the negotiation and decides what to promise a client.

This guide explains how each job is set up, where it helps and where it should stop. The examples come from systems we at ALTAI Digital built for estate agents in Istanbul.

Which jobs can AI take off an estate agent's desk?

1. First reply: questions from WhatsApp, Instagram and the website form get a reply drafted from information the agent has already written down.
2. Listing copy: the property details turn into a draft description, which the agent edits and approves.
3. Photos and video: listing photos are retouched within honest limits and short promotional clips are cut.
4. Client records: each client's budget, preferred areas and past conversations sit in one record.
5. Matching: when a new listing arrives, the clients whose wish list fits it are listed.
6. Land registry data: parcel information is reached from the same system.
7. Search visibility: the agent shows up on Google and in the answers of AI assistants.

What these jobs have in common is repetition. An agent answers the same questions every day and writes a similar description for every new listing. Once the repetitive part is drafted with AI, the agent's time goes to viewings, negotiation and completion.

How does AI help an agent reply to enquiries faster?

In property, clients often skip the contact form. We saw this with the Kali team in Zekeriyakoy and Sariyer within two weeks of launching their site. Buyers browsed the listing pages, then closed the browser and wrote on Instagram or WhatsApp. The agent was following three apps and losing track of which conversation was about which property.

The fix is a single queue. WhatsApp, Instagram, the web form and the phone all land in one panel. Common questions about floor area, service charges, floor and viewing times get a reply drafted from the agent's own information. Meta's AI feature in the WhatsApp Business app can also be set up for this, and the agent can take over the conversation at any moment.

Three rules do not change:

• The client is told plainly that they are writing to an AI assistant.
• If the answer is not in the agent's information, nothing is invented and the conversation passes to the agent.
• Price negotiation and contract questions always go to the agent.

Speed matters even more for advertising leads. In the flows we build, the name and phone number from an ad form reach the sales contact within seconds, by Telegram, WhatsApp, the panel or email. We describe the channel setup on our WhatsApp automation and Instagram DM automation pages.

How do you write a property listing with AI?

Start with the facts: floor area, number of rooms, floor, aspect, shared facilities, transport and the agent's own notes. A draft is produced from these facts, and the same facts can produce drafts in several languages. The agent reads the draft, corrects it and approves it.

Two things need care. The first is accuracy. A feature that is not visible in the photos, or that the agent has not checked, does not go into the text. Phrases such as “sea view” or “new build” go in only when they are true.

The second is authority. In Turkey, online property listings are verified through the Ministry of Trade's Electronic Listing Verification System (EIDS). According to the Ministry, authority checks began on 15 September 2024, and an agency may list a property only after the owner authorises it through the e-Devlet government portal. AI does not replace that step, and every market has its own version of these rules.

Once the listing details are prepared in a panel, sending them to the portals gets easier too. In our own product, AGENTSUP, listing details are prepared in the panel and pushed to the portal from the agent's own account with a browser extension.

Can AI be used for listing photos and video?

Yes, as long as the property stays true to life. The Kali system has a separate studio section for listing photos and videos, and Instagram and the other social accounts are run from the same panel. Tools of this kind correct the light in a photo and prepare short video edits and post captions.

The limits are simple. A crack in the wall is not removed, a view is not added and a room is not made to look larger. If virtual furniture is placed in an empty room, the image says so. When a buyer comes to a viewing, they should find what they saw in the photo, or trust is lost at the door.

How do you set up client records and listing matching?

AI only helps when the records are in order. If they are scattered, even a good tool gives poor results. So the first job is to move what sits in phones, spreadsheets and old portals into one place without losing anything.

We did this for Mujgan Simsek. 2,887 contacts were exported from her Keller Williams archive and 2,826 records were matched by phone number and enriched. The migration finished with zero errors, and both the raw backup and the rollback file were kept. The details are in the Mujgan Simsek case study.

At Kali, listings, enquiries and client records live in the same system. When we measured it for the case study, it held 1,138 client records, 94 listings and 431 images. When a property link is shared in a conversation, that conversation is attached to the listing's record automatically, so it is clear which client is interested in which home. The full build is in the Kali case study.

With clean records, matching can be added. When a new listing is entered, the clients whose budget, area and room count fit are listed. A message draft is prepared for each of them, and the agent decides who receives it. See our estate agent CRM page for the setup.

Can land registry data sit on the same screen?

It can. In Turkey, the Land Registry and Cadastre's Parsel Sorgu service shows parcel information to the public. In the Kali system, land registry and cadastre information is reached from the same system, so the agent does not search for it in a separate app.

This is a first check, not a legal search. Charges, restrictions and mortgages appear on the official title record, which is consulted before anything is promised to a buyer.

How does an estate agent show up in AI search?

A buyer may now type “agent who sells villas in Zekeriyakoy” into Google or ask ChatGPT the same thing. Showing up on both rests on the same base: a fast site, area pages, content that answers the question directly and a complete Google Business Profile.

We built the Koy Zekeriyakoy site for the estate agent Kerem Ali Yorukoglu. The site received SEO, GEO and AEO work and gained visibility in AI search. It is linked to his main site, so an enquiry does not wait on any channel, and we maintain it.

In our own measurement on 23 September 2026, ChatGPT answered part of the local business questions with a map card that showed ratings and review counts. Google also says that more reviews and positive ratings can help local ranking. So the Business Profile and genuine reviews are part of the work. We explain GEO and AEO in our guide to GEO and AEO and the service itself on our AI SEO page.

What should be in place before AI runs your property ads?

Measurement. Meta and Google adjust targeting and bids with their own AI, but those systems cannot learn without clean data.

We ran a Meta campaign for the Kali team between 29 August and 14 September 2026. The pixel, server-side events and lead matching were connected before the campaign started. The first ten days were expensive, and we saw that in the numbers rather than guessing it. After the form was changed, the cost per lead fell from 1,767 TL to 587 TL. In the villa campaign the same cost fell from 2,927 TL to 518 TL. Every lead was written to the client record together with the ad it came from. The details are in the Kali ad campaign case study.

Where should AI not be used in real estate?

• Pricing and valuation: AI can list comparables, but the owner and the agent agree the price. A formal valuation is a licensed valuer's job.
• Negotiation and contracts: offers, deposits and contract terms stay with the agent. In AGENTSUP a contract is analysed clause by clause and a risk report is produced, but a person decides whether to sign.
• Licences and authority: agency licences and listing verification are the agent's responsibility.
• Personal data: phone numbers, budgets and messages are personal data under GDPR and Turkish data protection law. Bulk messages go only to people who opted in, and conversations are not used to train models.

Where should a small agency start?

1. Count for one week where enquiries come from.
2. Pick the single task that takes the most time. For the Kali team it was the first reply to incoming messages.
3. Gather the client list from phones and spreadsheets into one place.
4. Build only that task, measure it for a month, then move to the next one.

ALTAI Digital starts with a discovery of about a week. We map the agency's channels, listing routine and data, then prepare a demo. After your approval, every part we build joins one panel, so the whole agency runs from a single screen. We explain the method on our digital transformation and estate agency automation pages, and our own product AGENTSUP covers the property side from start to finish.