Conecta Imobi 2026: AI agents in real estate customer service and the race for the first few minutes

What was presented at the event

According to coverage by Brasil Inovador, many of the new products from the sector's technology companies involved AI agents:

• Dommus: two agents — Dom, calibrated to answer a customer's first contact in under five seconds and qualify them within five minutes, and Donna, focused on after-sales. A property management module is planned for the first quarter of 2027.
• Lastro: Lais, a conversational AI that re-engages stalled leads and works as a copilot for agents.
• CV CRM: version 8.0 of the CRM, with support for MCP (Model Context Protocol, an open standard for connecting AI models to systems), so that agents can run automations with the customer's context.
• Wekoos.AI: an agent that works 24 hours a day by message in mortgage origination, compares rates from several banks and extracts data from documents automatically.
• UP Estate: guaranteed rent with tenant credit approval in about 30 seconds.

9 minutes versus 2 hours and 17

The most revealing figure at the event came from Imobilead, a company that diagnoses sales processes. According to an internal survey released by the company, well-structured agencies reply to a new contact in 9 minutes on average. Independent agents without technology support take about 2 hours and 17 minutes.

The same company says that customers contacted within 15 minutes are three to four times more likely to convert. These are figures from the company's own survey, with no published methodology, but they match the experience of any sales manager: a customer who messages three agencies tends to go with the first one that answers well.

That is why response time has become the main argument for AI agents in the sector. Answering in seconds, at any hour, is exactly what a team cannot do on its own — especially outside business hours.

AI as the agent's assistant, not a replacement

In the CNN Brasil report, Thiago Ely, MRV's vice president of sales and marketing, summed up the role the homebuilder sees for the technology: "Artificial intelligence is very much an assistant to the agent". According to MRV data, buying a property in Brazil takes 13 to 15 months on average — a long cycle in which a contact is easily lost if no one follows up.

Ely mentioned using AI to split the lead base into smaller groups and send targeted actions to each one, which, he said, makes it possible to "monitor much better". He also highlighted support for training agents, many of whom come from other fields and have not yet mastered topics such as mortgage credit.

Olivier Aizac, CEO of Grupo OLX, described uses on the portal side, which lists about 6 million properties: a virtual agent that answers customers 24 hours a day, price suggestions for agents based on the regional average, and analysis of search patterns to show agencies what kind of property people are looking for.

Where implementation works — or stalls

The products launched at the event have one thing in common: an agent is worth what it can look up and record. For anyone considering putting AI into customer service, these are the points to watch:

• Integration with inventory and CRM: an agent that does not check real availability will offer properties that have already been sold or rented. Access to data comes before the agent.
• Clear rules on what it can say: list price and features, yes; discounts, special terms and promises of credit approval, no. Financing simulations always appear as estimates.
• Handoff to the agent at the right moment: when the customer asks to negotiate, wants a visit or shows urgency, the agent takes over — with a summary of the conversation, without making the customer repeat everything.
• Transparent lead distribution: the AI agent qualifies, but the rule for distributing leads among agents (the "rotation") must remain clear to the team.
• Transparency and data protection (LGPD): let customers know that the first contact is handled by AI, ask only for the data you need and record everything in the CRM, with a defined purpose.
• Measurement from day one: time to first response, visit booking rate and conversion by source. Without this, there is no way to know whether the agent is helping.

Frequently asked questions

Does an AI agent replace the real estate agent?

No. The AI agent handles first contact, qualification, scheduling and follow-up throughout a buying cycle that can last more than a year. Visits, negotiation, document review and closing stay with the real estate agent — who starts receiving better-qualified contacts with context.

Do I need to change CRMs to use an AI agent?

Usually not. What matters is that the CRM and the inventory system allow integration, through an API or features such as MCP. When the system offers no integration, there are still alternatives, but the initial scope tends to be narrower.

Can the AI agent run financing simulations?

It can collect income, down payment and term and present an estimate based on the tables and rules provided by the company or the bank correspondent. Credit approval always belongs to the bank, and the agent should make that clear so as not to create false expectations.