AI real estate agent, the first deal has been closed.
On October 15, Ms. Wang, the seller, and Mr. Liu, the buyer, officially signed the sales contract for a second-hand apartment in Phase 6 of Zhuhai Huafa New Town, and then smoothly completed the property transfer procedures. This housing transaction with a total price of 1.95 million yuan is the first deal in China led by an AI real estate agent.
All core transaction links of this deal, including online listing, customer matching, housing source authenticity verification, output of community transaction reference price, and online price negotiation, are provided by the AI agent "E-House Xiaoxin", while human practitioners only participate in links such as offline on-site viewing, professional contract notarization, and transfer service.
Another special point of this transaction is that the AI agent does not charge transaction commission. The seller gave all the 40,000 yuan commission that would have been paid to the intermediary under the traditional mode to the next owner of the house as a full benefit transfer.
From the "auxiliary" role to truly standing at the core of transaction matchmaking, AI has finally taken a key step.
At present, the AI applications in China's real estate brokerage field are divided into two completely different paths. The mainstream mode takes AI as a tool to improve efficiency and service, assisting agents in their daily work. AI is an add-on for intermediaries, which does not change the original business model, keeps the commission rules unchanged, and the transaction decision-making power remains in the hands of agents. Platforms including Beike, 58 Anjuke, and small and medium-sized intermediary platforms are using it on a large scale.
Representative products include Beike AI Workbench, Anjuke Intelligent Chat, various third-party real estate copywriting/digital human tools, etc. At present, there are corresponding application tools in links such as housing source authenticity verification, information structuring, valuation, housing content production, hosted customer service, lead scoring, conversion prediction, and tax calculation. They are characterized by maturity and large quantity, but AI does not independently connect buyers and sellers, and cannot complete matchmaking without agents.
The other type is AI independent intermediary Agent. AI is no longer a tool for agents, but an independent and neutral intermediary role with independent matchmaking authority, which can initiate conversations between buyers and sellers without relaying messages through agents. Its goal is to directly complete supply-demand matching, two-sided matchmaking, preside over price negotiation, and realize direct connection between buyers and sellers. However, this type of mode is rarely applied at present, and is still in the exploration period of forming a complete closed loop of transaction process and commercial path.
There is another biggest difference between the two modes: the AI in the first mode mainly serves intermediaries such as real estate agents, while the second mode directly provides services to C-end individual property owners and individual home buyers.
Different from new properties, the buyers and sellers of second-hand housing are mostly scattered individuals, and transactions rely more on intermediary services. Therefore, the transparency and commission rate of second-hand housing brokerage services have always attracted much attention. Price suppression, price raising and concealment of transaction information by agents often lead to transaction frictions. The professional knowledge reserve of agents and their familiarity with housing sources also directly affect the transaction service experience.
As a result, the demand for AI in the real estate brokerage field is growing increasingly strong. According to the statistics of *2026 Survival Report for Millions of Real Estate Agents* released by 58.com Anjuke, the AI utilization rate of national real estate agents reaches 49.7%, and 58.9% in first-tier cities, which is mainly used in scenarios such as housing source promotion, collection of commercial district and real estate information, analysis of customer needs and matching of housing sources. The report believes that agents' demand for AI has shifted to "whether it can solve practical business problems".
Completely different from the path of AI empowering agents, the AI intermediary intelligent agent represented by E-House Xiaoxin tries to restructure "intermediary services", and its application of AI is more radical. It separates the intellectual intermediary work in real estate brokerage, such as "information matching, market interpretation, and process organization", from the commission package, and all of them are undertaken by AI.
Specifically, the AI agent mode of E-House Xiaoxin divides the entire intermediary service into two layers: the four high-value intellectual work of information, matching, matchmaking and recommendation in the intelligent decision-making layer are completed by the AI agent for free and in a neutral manner; the errand execution work in the performance execution layer, such as viewing, mortgage, transfer, and house inspection, are separated as standardized professional services, which are clearly priced, provided on demand and charged according to actual workload by special performance service providers.
The whole set of mode realizes direct dialogue between buyers and sellers, AI undertakes intermediary functions, human service companions follow up the whole process, and a professional performance team is responsible for landing implementation. The core breakthrough of the mode lies in reducing the transaction cost of the intermediary link to zero. The price negotiation concessions of buyers and sellers can be directly converted into their actual benefits.
From the perspective of resource allocation, this split brings a welfare improvement: after zero commission and transparent performance fees, intermediaries have no motivation to pollute information or distort matching, so the information is more real, the matching is more in line with actual demands, and the matchmaking is more fair. While the total transaction cost decreases, information asymmetry is reduced, and resource allocation efficiency is improved accordingly.
The logical support for AI to act as a real estate intermediary lies in two aspects: first, the information processing capability. AI can efficiently complete the structuring of housing source information, authenticity verification, and market data comparison, reducing artificial information asymmetry. Moreover, AI has no interest demand for closing a deal, so it will be more neutral. The AI intelligent agent does not depend on the commission for survival, so it will not hide the defects of the property to facilitate the transaction, publish false information to grab customers, or favor one side in the game; second, the marginal cost of AI is almost zero, so it can completely deconstruct the "high cost" of intermediary fees. The cost for an intelligent agent to serve 10,000 customers at the same time is almost the same as serving one customer. This means that the most valuable intellectual work of intermediation (information, matching, matchmaking, service recommendation) can gradually tend to be marginally free.
The first transaction in Zhuhai proves that the link of AI real estate agent can operate smoothly under ideal conditions, but up to now, the publicly available AI intermediary transaction cases in China that have completely gone through the full link of matching - price negotiation - property transfer are still very rare. It is more like an industry paradigm sample, proving that the traditional human-centered role of real estate transaction intermediary can be replaced by Agent intelligent agents. However, whether it can be replicated on a large scale depends on three variables: the capability of localized performance network, the willingness of retail owners to list their properties, and the market's risk acceptance of "no-intermediary direct transaction".
There are still many problems to be explored and solved in the new mode. For example, real estate transactions involve large amounts of money and long chains. How to define the responsibility when disputes occur in AI matchmaking, how to ensure the quality of offline performance services, and how to price the service charges beyond zero commission in the long term, all need to be tested in more actual transactions.
AI is changing real estate brokerage, but it will not simply "eliminate intermediaries". The application of AI in this track is still mainly focused on short-term efficiency improvement, and the long-term intermediary reform is advancing slowly. The path of AI independent intermediary Agent will continue to carry out pilot trials and errors in some cities. Its real value is not to seize large market share in the short term, but to provide the whole industry with a brand new possibility.