Since August 11, 2026, real estate agencies in France can no longer contact individuals by phone without their prior agreement, as mandated by Law No. 2025-594 of June 30, 2025. This law requires that any phone contact with an individual must be preceded by their explicit consent, and the proof of this consent must be kept by the agency. Decree No. 2026-662 of July 23, 2026 clarifies that these agreements are valid for a maximum of one year. Phone calls are allowed from Monday to Friday, excluding holidays, between 10 a.m. and 1 p.m. and from 2 p.m. to 8 p.m., with a maximum of four attempts over thirty days. An exception exists for clients who are already under a contract related to the subject of the call. For real estate agencies, this means that individuals who list their properties can no longer be contacted by phone unless they have given their consent beforehand. As a result, individuals must either visit the agency in person or have previously agreed to be contacted, often through the agency’s website. Many agencies are exploring tools to replace the traditional phone call, which was previously used to schedule property appraisals. The core tasks remain: identifying potential clients, sorting requests, estimating property values, approaching clients, and reporting updates. AI is being used to streamline these tasks. Experts recommend assigning one agent per task rather than relying on a single universal assistant. A conversational assistant is designed to support a human negotiator who knows how to guide it. It waits for input from the user. In contrast, an AI agent functions with a specific mission, input data, an expected result, and a stopping rule. It operates independently, often during off-hours, and is connected to the agency’s transaction software when its interface allows. Five AI agents are used to cover the essential parts of the preparation process. The first AI agent prepares outreach that is still permitted. It works from the agency’s files, including previous clients and those who have already given consent. It also considers public data, such as energy audits published by ADEME, which often precede property sales. It identifies individual listings and prepares a letter addressed to the property, which the negotiator reviews and signs. It does not initiate any calls. The second agent sorts incoming requests. It processes appraisal forms, messages from online portals, and visit requests, distinguishing between sellers and buyers, noting the project, deadline, and area, and preparing a response. In the morning, the negotiator finds a prioritized list rather than an inbox full of messages. The third agent prepares property appraisals. It uses the property’s characteristics and comparable sales in the neighborhood to propose a reasonable price range. The negotiator reviews and adjusts this before sending it. An online appraisal is often the first point of contact for a seller, so it is crucial to respond quickly and accurately. The fourth agent connects properties with potential buyers. It compares new mandates with existing buyer files and matches new buyers with available properties. The negotiator receives these matches with explanations and decides whom to contact. The fifth agent updates sellers. It compiles information about visits, buyer feedback, and competing listings in the neighborhood to prepare regular updates that sellers expect. The AI agent makes the decision, while a standard automation executes the task. A common mistake is to ask the AI to handle everything, including sending messages. Instead, it is better to separate the roles: the AI agent analyzes and decides, while a standard automation handles the execution. This separation makes the system more controllable and ensures compliance. The automation knows when to stop, such as avoiding calls without recorded consent or outside permitted hours. The AI prepares, and the negotiator validates. The law places the burden of proof on the professional, and the tool helps maintain compliance. Each agreement collected on the agency’s website is timestamped, attached to the contact file, and marked with an expiration date. The agent alerts the team when agreements are about to expire. The decision to send a letter or appraisal remains human. The negotiator no longer spends their morning searching for whom to call but instead chooses whom to meet. The management approach has also changed. Previously, a real estate agency director might have tracked the number of calls, but now they can monitor incoming requests, response times, appraisals sent, and appointments made. Starting with sorting incoming requests is a good initial step, as it does not send anything on its own and is easily measured by response time and the number of appointments made. Two conditions are more important than the tool itself: a transaction software that allows reading and writing data and a website that attracts requests, such as neighborhood pages, online appraisals, customer reviews, and agreement collection. Without incoming requests, there is nothing to sort. August 11 has not eliminated the need for sellers, but it has shifted the effort from the phone to the website that attracts sellers and the organization that responds to them. The AI agents do not replace the negotiator but return the time that used to be spent on outreach.