In short

AI for real estate agencies prepares and remembers: market studies, seller reports and CRM notes. The conversation with the seller, the listing appointment, the negotiation and coaching your team stay with people. Start with one process that already works and is written down. And since August 2026, anyone talking to an AI assistant in the EU must be told.

Having ChatGPT on every agent's phone is not enough to make your agency work with AI. I see it in many of the agencies I work with: everyone uses it for something different, and the owner couldn't tell me a single task that AI handles from start to finish.

The tools are already in the office. According to Spain's national statistics office (INE), 33.6% of Spanish companies with 10 or more employees used artificial intelligence in the first quarter of 2026, against 21.1% a year earlier. The INE warns that, after a change of classification, the two years are not strictly comparable. Among companies with fewer than 10 employees and an internet connection, the figure was 21.4%. What's missing is deciding which work you hand over.

To me, AI for real estate agencies means giving an assistant complete tasks that a person on your team does today, with your instructions and your data, and measuring whether it gives you hours back. That is what this guide covers, ending with what the EU AI Act asks of you.

‘AI prepares. The professional talks. The system remembers.’

Michele Di Sei, coach for real estate leaders

 

That line sums up the method. Everything below comes from it.

AI for real estate agencies: which tasks can it handle?

The ones that prepare or record the work: the market study for the listing appointment, the Friday report, viewing reports, CRM notes, reminders and first drafts. The ones that decide or persuade stay with a person. Here are the real estate tasks to automate with AI, and the ones to keep, as I set them out for agency owners:

Task Hand it to AI? Who does what
Preparing the listing appointment: market study with comparables and a draft prelisting pack Yes AI cross-checks portal comparables and sold properties from your CRM. You check the price and defend it in front of the seller. What happens next is in the data on winning exclusive listings.
The Friday report to the seller Yes AI drafts it from the week's numbers: enquiries, viewings, second viewings and offers. You make the Friday call yourself.
Viewing reports and purchase offers Yes They come out of your CRM data. You check them before anyone signs.
Call notes and summaries in the CRM Yes You dictate after hanging up and AI tidies it up. You decide the next step.
Reminders and follow-up cadence (36 touches a year) Yes AI reminds you and drafts the 26 fortnightly newsletters. The 4 calls of the year, one per quarter and with a real reason, are yours.
First drafts of property ads and copy Yes AI drafts, a person reviews: no property detail goes out unchecked.
Planning the week's diary Yes It suggests the order. You protect the hours for conversations.
Talking to the seller and the listing appointment No That is where they decide whether to trust you.
Negotiating price and fees No If it goes wrong, your name is on it.
Hiring or letting people go No It is a decision about people. Using AI to screen candidates or assess workers' performance may count as high risk under the EU AI Act.
Coaching your agents No Agents change when someone looks at their work every week and tells them face to face.
Marketing emails to people who never asked for them No one, AI included In Spain, the LSSI (art. 21) bans them without prior consent, except to existing customers under its conditions.

 

Look at the ‘yes’ rows: AI delivers a draft and a person reviews it before it leaves the agency. If the AI gets something wrong, the mistake stays in-house.

What should AI never do in a brokerage?

It should never hold the conversation that decides the deal. Property deals are closed in conversations, and those conversations can't be delegated.

A seller hesitating over an exclusive listing is afraid of selling too cheaply or picking the wrong agency. You don't sell flats. You defuse fears. That takes someone sitting in their living room, hearing what they don't say and holding the silence. AI can help you prepare that appointment; you're the one who goes.

Negotiation is the same, for a different reason: price and fees need judgement and someone who answers for it. If the seller loses a good offer because a tool told them to hold out, the name they'll remember is yours.

You can't hand coaching your team to an assistant either. It can prepare an agent's numbers before Monday's meeting. The awkward conversation, the ‘three weeks without an appointment’ one, is yours. I wrote about it in why AI can't think for you as a leader.

The test I put any tool through fits in one question: does this AI lead you to more conversations with potential clients? If not, AI isn't your problem.

Where should I start using AI in my agency?

With sales. The order I follow with agency owners is sales, then systems, then AI, and I don't change it however rushed the team is. Five steps:

  1. Make sure your team can sell and communicate. According to the MoneyMaker 2026 Report (Informe MoneyMaker 2026: 30 agents from different groups, 2023-2026, each measured over their own 12 weeks), 4,514 conversations led to 417 listing appointments, 241 exclusive listings and 198 transactions, sales and lettings included. It all starts with a conversation. If your people aren't having them, AI will just do unproductive work faster. More in why AI won't save your agency.
  2. Pick one process you already run and write it down. Just one, the Friday report for instance, on a single page: the steps and who does them. What lives only in your head can't be automated, and the cost of not having a system is something I've covered separately.
  3. Automate it with AI and measure it. Before you touch anything, time three reports this week. Write the instruction with RCCF (role, context, command and format), test it on real data and time it again. Without that measurement you won't know if it worked, and that is what I see behind so many failed AI initiatives in agencies.
  4. Repeat with the next one. Move to the second process once the first runs without you watching over it. Two half-done processes are worth less than one finished.
  5. Delegate the 80% that doesn't produce so you can focus on the 20% that does, which is conversations and your team. To decide what to let go of first, use the delegation matrix for agency owners.

A note on the Report: the data is self-reported by the agents and comes from several Spanish-speaking markets, not only Spain.

What is a team of AI agents for a real estate agency?

An AI agent is an assistant that does a complete task for you, following your instructions and your data: for example, preparing the market study for a property or writing the weekly report to the seller. A team of AI agents is the same idea, one per process, all working with your agency's information.

Three of my students' cases, with no names and no client data:

  • One has built an assistant that prepares the market study before every listing appointment, using portal comparables and sold properties from the CRM.
  • Another produces viewing reports and purchase offers straight from CRM data.
  • A third has the Friday seller report scheduled to go out automatically.

That is what AI agents for real estate look like when they work. None of the three is a programmer. Before AI, they already had a process they did by hand and could explain.

That is what's new in MoneyMaker 3.0. MoneyMaker is a 12-week training programme that teaches you to sell and to systematise your work. Version 3.0 adds one step after that: building your agency's team of AI agents yourself, with no technical knowledge. The order stays the same: sales first. One caveat: you pay for an AI subscription, but the processes, the instructions and the data are yours. The details are on the MoneyMaker programme page.

What does the EU AI Act mean for real estate agencies?

For an agency operating in the EU, two articles matter most today: Article 4, on training your team, and Article 50, on telling clients they are talking to an AI. This is what the official text says, as checked in November 2026:

  • The EU AI Act is Regulation (EU) 2024/1689. It applies in phases, and its general date of application is 2 August 2026.
  • AI literacy (Article 4), applicable since 2 February 2025. The AI omnibus, Regulation (EU) 2026/1744, published on 24 July 2026, rewrote it: it now asks you to take measures that support the AI literacy of the staff using these systems, without requiring you to guarantee a specific level for any individual. A simple way to start: put in writing which tools your team uses and which data never goes into them.
  • Transparency (Article 50). Since 2 August 2026, an AI assistant that talks to people, on your website or on WhatsApp, must be designed so that people know they are dealing with an AI, at the latest at the first interaction, unless it is obvious. The obligation falls on the provider, meaning whoever develops the assistant and puts it into service under their own name or trade mark. If you build it yourself for your agency, that provider may be you: check with your adviser. Saying it is an AI in the first message costs you nothing.
  • Decisions about your team. AI systems used to select candidates or evaluate workers' performance are listed as high risk (Annex III, point 4). After the omnibus, those obligations apply from 2 December 2027.
  • Personal data of sellers and buyers: the GDPR still governs it, together with national law (in Spain, the LOPDGDD and the Spanish Data Protection Agency, AEPD). Never put an ID document into a tool if you don't know where it stores the data.
  • In Spain, AESIA (the Spanish Agency for the Supervision of Artificial Intelligence) describes itself as the public body responsible for ensuring the ethical and safe use of AI.

This guide is general information, not legal advice. If you work outside Spain or the EU, or have a specific case in front of you, talk to your adviser.

What AI training do agency owners and managers need?

More judgement than tool tutorials. From what I see, a lot of AI training for real estate agents teaches which button to press. An owner needs to know which process in the agency deserves automating first and how to measure whether it worked. The button takes an afternoon to learn. Choosing the right process is what separates agencies that win back hours from those paying for subscriptions nobody opens.

Under Article 4, making sure your team knows how to use AI is also a company obligation. How it fits with everything else a manager needs to learn is in training for real estate agency managers.

What do you do this week?

Pick a process, the Friday report if in doubt, and time three of them. By next Monday you'll have your agency's first AI number, and it will be yours.

Book your strategy call (45 min) if you'd like to talk it through with me. And if you're thinking about working with someone from outside, read what a real estate coach does and how to choose one first.

Frequently asked questions

Which real estate agency tasks can be delegated to AI?

The ones that prepare or record the work: the market study for the listing appointment, the weekly report to the seller, viewing reports and purchase offers, call notes in the CRM, follow-up reminders and first drafts of property ads. Always with a person reviewing before anything goes out. The conversation, the negotiation and coaching the team stay with people.

What should AI not do in a real estate agency?

It should not hold the conversation with the seller or the listing appointment, negotiate price or fees, decide whom to hire or let go, or coach your agents. Trust and responsibility are at stake there. AI can prepare each of those conversations. Having them is the job of a person in your agency.

Where should I start using AI in my agency?

With sales, before any tool. First, make sure your team can sell and communicate. Then pick one process you already run, such as the Friday report, and write it down on one page. Next, automate it with AI and time it before and after. Once it runs without you, move to the next one. Finally, delegate the 80% that doesn't produce.

What does the EU AI Act require from a real estate agency?

Regulation (EU) 2024/1689 asks you to take measures that support your team's AI literacy (Article 4, amended in July 2026) and, since 2 August 2026, that an AI assistant talking to clients tells them it is an AI, unless that is obvious (Article 50). Using AI to select or evaluate staff can be high risk, with obligations from December 2027. Personal data still falls under the GDPR. Check your case with your adviser.

Can I build AI agents for my agency without coding?

Yes. There are agency owners and agents already preparing market studies, viewing reports and seller reports with assistants they built without being programmers. What you need first is a process that already works and that you can explain. In MoneyMaker 3.0, after sales and systems, students build their agency's team of AI agents themselves, with no technical knowledge.

About the author

Michele Di Sei is a coach for real estate leaders. He has worked in real estate since 2012 and has trained more than 500 agents and managers through courses, webinars and programmes. He built and led a team at Keller Williams Marbella and was a trainer for Keller Williams Spain. He created MoneyMaker and MoneyMaker Leader and has spoken at AMPSI, RE/MAX (2026), Inmociónate, the Agora Congress, Family Reunion Spain and Keller Williams Mega Camp in the United States.

 

LinkedIn · YouTube

Sources