AI Agents for Real Estate: Buyer Qualification at Every Hour

Direct answer: An AI real estate agent, deployed on new-build developer sites as an AI concierge for off-plan property sales, is a face-and-voice system deployed on property portals, estate agency websites, or developer sites. It answers buyer questions about listings, qualifies buying position (budget, timeline, mortgage status), books viewings, and routes motivated buyers to a human agent, all at any hour, in any language, without a human agent needing to be available.

Property searches happen at night. A buyer who has been thinking about a specific neighbourhood for months finally finds the listing that fits. It is 10:47pm. The estate agency is closed. The "request a viewing" form leads to a two-day wait. The buyer fills it in, briefly, and by the time the agent calls back, the buyer has moved through three more portals and the sense of urgency has passed.

The buying cycle in residential property is long, six to twelve months for a typical UK or EU buyer, but the emotional intensity is episodic. The moment a listing resonates, the buyer wants to engage now. A Human AI agent that is present at that moment, fields every question, and books the viewing in the same conversation saves the agent a follow-up call and captures a motivated buyer at peak intent.

What buyers ask at 11pm

The questions are predictable. What is the service charge? Does the lease have ground rent? Is the area flood-mapped? What does the planning history look like? How long has it been on the market and why? Are the sellers in a chain? When can we see it?

None of these questions require a human agent to answer. They require information, structured, accurate information about the property, the legal status, the local area, and the process. A Human AI agent configured with that information answers all of them, in the buyer's language, at whatever hour the buyer is ready to ask.

At the end of that conversation, the agent knows: budget confirmed, mortgage in principle held or not, property type preference, chain status, and ideal viewing window. The human agent's first contact is with a qualified, briefed buyer, not a cold enquiry from a form. Speed matters here more than most agents assume: Rightmove has publicly urged agents to respond within 24 hours or risk losing the lead, and independent surveys have found nearly 40% of agents taking longer than that to reply to a portal enquiry, exactly the window in which a buyer's interest cools or moves to a competing listing.

New developments and off-plan sales

For new build developers and off-plan property sales, the Human AI agent use case is especially strong. Development websites receive high-intent traffic from buyers researching before the show home opens. Questions are specific: What is the build programme? What are the specifications for the kitchen? Is there parking? What is the estimated service charge? Are there any remaining units on the south side?

An AI agent on a development website that handles these questions in a live, face-to-face conversation, and then reserves a show home appointment or adds the buyer to the priority list, converts at a higher rate than a brochure download form. The buyer engaged. The buyer talked. The reservation follows naturally. There is no single published industry figure for online-to-reservation conversion specifically, but the Home Builders Federation's National New Homes Customer Satisfaction Survey shows the downstream effect of getting this relationship right from the outset: 93% of new build buyers say they would recommend their builder, and more than nine in ten would buy new build again.

International and multilingual buyers

In major EU property markets, Berlin, Amsterdam, Paris, Lisbon, Madrid, a meaningful proportion of buyers are non-native speakers. A Human AI agent in the buyer's language removes the friction that causes multilingual buyers to abandon a local property portal in favour of international-facing alternatives. The voice layer supports multiple languages, giving EU real estate markets solid coverage across the region's main buyer languages.

The compliance boundary: information vs advice

Real estate AI agents operate under the same advice boundary as financial services agents. In the UK, estate agents are regulated under the Estate Agents Act 1979 and Consumer Protection from Unfair Trading Regulations 2008. In the EU, property transaction intermediaries fall under national consumer protection law and, where mortgage advice is involved, MCD (Mortgage Credit Directive).

The AI agent handles property information and process explanation. It does not provide mortgage advice, valuation opinions on whether the asking price is fair, or legal guidance on contract terms. These are human professional responsibilities. The agent is transparent about this boundary and routes mortgage, legal, and valuation questions to the appropriate human or professional service. As the House of Commons Library's briefing on estate agent regulation sets out, simply providing information or advertising a property does not in itself constitute regulated estate agency work, but negotiating on a client's behalf does, which is the line an AI agent's configuration needs to respect. Holding that line inside a live conversation, where a buyer's follow-up question slides from information into advice without anyone announcing it, is the hard part of this deployment, and it is where Ojin puts most of its effort.

Ojin builds the face-and-voice layer for this, and the platform is self-serve, so an agency can put one agent on one branch's portal feed and compare it against the branch next door before rolling anything out. That is a cheaper experiment than any portal upsell.

When the enquiries actually arrive

The timing problem in estate agency is unusually well documented, and it is the whole argument for an always-available agent. Yomdel's work on lead response in property reports average response times to portal enquiries measured in hours rather than minutes, with a large share of enquiries arriving in the evening, after offices have closed and before anyone is back at a desk.

Put those two facts together and the gap is obvious: the highest-intent moment in the funnel lands at 9pm and gets answered at 10am the next working day, if at all. The general speed-to-lead research says the same thing in a different market. The MIT and InsideSales study published in Harvard Business Review found the odds of qualifying a lead drop sharply between a five-minute and a thirty-minute response, and a property enquiry is not more patient than a software one. An Ojin agent answers at the moment the enquiry lands, which in this market is usually the evening.

What the agent should not be allowed to say

Property is a regulated sale in most European markets, and the boundary matters more than in retail.

An agent can describe a property, explain the buying process, set out what documents a buyer will need, book a viewing and answer questions about a development's timeline. It should not value a property, advise on whether a price is fair, comment on mortgage suitability, or make representations about a building's condition or legal title. Those are either regulated activities or statements a buyer could later rely on in a dispute.

The practical design is the same as in financial services: the boundary lives in the knowledge base rather than in the prompt alone, and questions that approach it escalate to a named person rather than being answered carefully.

Where it pays for itself

Three places, in the order most agencies see them.

Out-of-hours capture, which is the largest and the easiest to measure, because the counterfactual is a voicemail.

Qualification before viewing, which protects the most expensive resource in the business. A viewing arranged for someone who cannot finance the purchase costs a negotiator an afternoon and a vendor their patience.

Cross-border enquiries, where the alternative is a translated email chain running over several days and usually ending in silence.

Frequently asked questions

Where does Ojin sit in a deployment like this?

As the technology provider and, under GDPR, the data processor. The agency remains the controller and the regulated party, and Ojin runs EU-based infrastructure from Berlin.

Can the AI agent handle leasehold or commonhold queries?

It can explain what leasehold and commonhold are, what service charges and ground rent mean, and what the buyer should ask a solicitor. For specific advice on a particular lease's terms, it routes to a solicitor. Information delivery yes; legal interpretation no.

How does the agent handle a buyer who wants to negotiate price?

It acknowledges the buyer's position and connects them with the human agent who handles price negotiation. "I can pass your position to [Agent Name], they'll be in touch within [timeframe] to discuss." The conversation context (buyer's budget, motivation, and chain status) is included in the handoff.

Can the AI agent book viewings directly into the agent's calendar?

With calendar API integration (Google Calendar, Outlook, or estate agency CRM like Reapit or Dezrez), yes. The agent surfaces available viewing slots and confirms the booking within the conversation. Integration documentation at docs.ojin.ai.

Next steps for a real estate deployment

For the neighbouring verticals, see AI agents for financial services, which shares the regulated-advice boundary a mortgage conversation runs into, and AI agents for retail for the same high-consideration guidance problem on a shop floor. For the sales motion, see what an AI sales agent is and how it differs from an AI SDR and how a Human AI Agent handles sales conversations. For the category, see what a Human AI Agent is and how it actually works and what a conversational AI agent is and how it differs from a chatbot. Property portals, off-plan sales, multilingual buyers, the compliance detail and the virtual tour comparison each get their own article in this vertical.

Enterprise and developer deployment: ojin.ai/enterprise. Platform demo: try a live agent.