The landscape of AI in the real estate industry has changed.
In 2025, with the emergence of DeepSeek, the once-popular "Hundred Models Battle" has completely fizzled out. Only a handful of "deep-pocketed" giants are still qualified to stay at the table of the foundation large model game.
In contrast, there has been an explosion of Agents. Rumors have it that Manus' ARR is approaching $100 million, and its latest valuation has soared to a high of $2 billion. In vertical fields such as Office work, scientific research literature, code programming, and even media podcasts, Agents have achieved explosive growth.
One of the most noteworthy AI Agents comes from an unexpected industry - real estate.
CRIC Deep Intelligence Connect Interface
In April this year, CRIC Group, a subsidiary of E-House (China) Holdings Limited, released China's first real estate AI Agent, "CRIC Deep Intelligence Connect," and opened it for internal testing. This is a professional Agent designed for real estate industry practitioners, with four major sections: search, articles, reports, and knowledge base. It can provide analysis, pricing, and marketing planning in areas such as cities, land, real estate projects, and rental prices.
On May 29th, after product upgrades and incorporating user feedback from internal testing, CRIC Deep Intelligence Connect was officially upgraded to CRIC Deep Intelligence Connect 2.0, transforming the AI Agent from an auxiliary tool to an output of professional industry results. At the same time, the first "New Real Estate Project AI Intelligent Agent" product for the marketing scenario of new real estate project sites was launched, enabling real estate users to conduct targeted AI analysis on areas such as the city, location, and competitors of new real estate projects - in the real estate industry, this may be one of the most important business scenarios for all enterprises.
On the same day, CRIC Group signed cooperation memorandums with four enterprises: Lianfa Group, China Jinmao, China Resources Land, and China Merchants Shekou. The two sides will jointly explore the in - depth integration and innovative models of AI technology in the real estate industry. Meanwhile, CRIC Deep Intelligence Connect 2.0 established the first batch of co - construction partnerships with 11 benchmark projects, focusing on the co - construction of the practical application of AI in the new real estate marketing scenario around the "New Real Estate Project AI Intelligent Agent."
It's easy to understand that CRIC Deep Intelligence Connect is the first AI Agent in the real estate industry.
But to some extent, it is also the first Agent in the real economy.
Currently, the large AI model industry has reached a critical industrial turning point, and the real economy's eagerness for industrial structural transformation and the leveraging of new - quality productivity is becoming more and more urgent. AI and Internet giants have also been coveting the "ToB Internet" for more than a decade.
An era of in - depth integration of AI and the real economy may be approaching.
The Real Economy: A "Hard Nut" That BAT Couldn't Crack in a Decade
According to the data from the "Research Report on China's Digital Economy Development (2024)" by the China Academy of Information and Communications Technology, in 2023, the scale of China's digital economy reached 53.9 trillion yuan, accounting for 42.8% of GDP.
Among them, the scale of digital industrialization in electronic information manufacturing, software industry, basic telecommunications, and Internet value - added services was 10.09 trillion yuan;
While the scale of industrial digitalization, which uses digital technology to integrate into the primary, secondary, and tertiary industries (such as agriculture, chemical industry, real estate, clothing, and transportation) and brings economic growth, was as high as 43.81 trillion yuan.
The internal structure of China's digital economy from 2012 - 2023. Image source: "Research Report on China's Digital Economy Development (2024)" by the China Academy of Information and Communications Technology
Yes, even when mobile phones and computers seem to have occupied every aspect of our lives, the real economy is still the absolute leader in terms of industrial scale.
This is why more than a decade ago, Internet giants such as BAT launched a high - profile offensive on "Industrial Internet," "Internet +," and "ToB Internet," claiming to "reshape everything" and that "all industries are worth doing over again." Calculated, the value created by signing one project is equivalent to selling 100,000 game skins, 1 million products at 9.9 yuan free shipping, or 10 million advertising clicks.
Unfortunately, from "Internet +" to "AI +," Internet giants have made some progress in the past decade, but not much.
In essence, the three most important elements in the real economy are:
1. Data; 2. Scenarios; 3. Customers.
First is data - or more precisely, industrial data.
The real - world industries are almost as old as human civilization. A large amount of industrial experience and pain points in scenarios are precipitated in construction sites and boiler factories, and cannot be found on the Internet or in PPTs.
What contemporary AI needs most is precisely industrial data.
With the full - scale open - sourcing of DeepSeek, the model capabilities of various Agents are quite similar. What really makes the difference is the training data that enterprises "pour" into them.
Take the real estate industry as an example.
Since 2024, at least 32 provinces and municipalities across the country have issued documents related to the pilot of spot - housing sales, promoting the spot - housing sales model.
This is a typical piece of news where "outsiders just watch the excitement, while insiders see the essence." If you are just an ordinary consumer and happen to look up some information after seeing this news, general large AI models on the Internet can answer most of your questions.
However, if you are a real estate developer with some development targets in mind and are considering participating in land bidding, it is definitely not enough for AI to simply answer that "the promotion of the spot - housing sales model will cause housing price fluctuations" or that "the sales area of spot - housing increased by 18% in 2023."
Analysis Report on the Impact of the National - Scale Spot - Housing Sales Pilot on the Real Estate Market from 2024 - 2025. The above picture was generated by CRIC Deep Intelligence Connect
vs
Analysis Report on the Impact of the National - Scale Spot - Housing Sales Pilot on the Real Estate Market from 2024 - 2025. The above picture was generated by other general AI Agents
As shown in the CRIC Deep Intelligence Connect report above, as a decision - maker, you hope that AI can, based on the latest policy trends and industry data in 2025, provide reports on the trend of land auction prices, the cost structure of development investment, and even environmental analysis reports of specific target real estate projects, and even directly give business suggestions...
Currently, large AI models generally use dual - training of "public data + private data." And the private - domain data of general industries cannot be found on the public Internet and are all precipitated in the enterprise's own database.
Zhou Xin, the Chairman of the Board of Directors of E - House China, once revealed that the private data of CRIC Deep Intelligence Connect mainly comes from CRIC's internal database, which covers professional data of 426 cities across the country, real - time market information, and real - estate knowledge graphs, including various sub - fields such as residential, land, office, commercial, rental, healthcare, industrial cities, cultural tourism, enterprises, and macro - policies.
It should be noted that E - House China has a 25 - year history since its establishment. It was listed on the New York Stock Exchange in 2007 and is the first Chinese light - asset real - estate concept stock listed in the United States. Its business has covered hundreds of cities across the country, and it has characteristic business segments such as the new - housing sales group, CRIC big - data group, and Leju Holdings.
25 years, 426 cities, covering a lot of data in residential, land, office, commercial, rental, healthcare, industrial cities, etc. - this is a highly vertical, highly accurate, and extremely commercially valuable "data gold mine."
These data were "collected" by real - estate people riding bicycles and electric vehicles, visiting one real - estate project after another and one office after another. Based on the accumulation of these data and industry knowledge, CRIC Deep Intelligence Connect is constantly evolving in terms of professionalism and depth. The following are excerpts from the Shanghai real - estate market monthly reports in April, which are the works of the ordinary version 1.0 and the advanced version 2.0 respectively. The improvement of version 2.0 can be seen even in the report overview section.
Excerpts from the "Shanghai Real - Estate Market Monthly Report in April." Left picture: Ordinary version of CRIC Deep Intelligence Connect; Right picture: Version 2.0 of CRIC Deep Intelligence Connect
Real - estate insiders will never judge the market trend based solely on public data such as the transaction volume and average price of new residential properties. They need first - hand industrial data from policies to land, from residential to second - hand housing and commercial properties, and need structured analysis and cross - comparison of different business forms, different regions, different area segments, and different price segments, such as the relationship between land supply and commercial - housing transactions, and the transaction structure of the area segment of commercial residential properties. Therefore, the professional version of CRIC Deep Intelligence Connect is still continuously working hard to output work results to meet the professional needs of industry practitioners.
In addition to data, in - depth cultivation of business scenarios is also of utmost importance.
Based on the in - depth integration of the real - world industries and AI, the idea of creating CRIC Deep Intelligence Connect Agent is "professional data + industry experience." Specifically, through a deep understanding of the real - estate industry, CRIC Deep Intelligence Connect refines, summarizes, and divides different business scenarios in the industry, and through in - depth exploration of the details of business scenarios and in - depth understanding of the pain points of industry users, it finally achieves AI services that are "useful, easy - to - use, and professional."
- Even for non - real - estate industries, this idea of creating a vertical - industry AI Agent is very valuable for reference.
Take the "New Real Estate Project AI Intelligent Agent" launched simultaneously with CRIC Deep Intelligence Connect 2.0 as an example:
CRIC Deep Intelligence Connect 2.0 New Real Estate Project AI Intelligent Agent. Image source: CRIC Deep Intelligence Connect
CRIC Deep Intelligence Connect 2.0 New Real Estate Project AI Intelligent Agent is the first AI Agent in the industry launched for the marketing scenario of new real - estate project sites. It is both professional and vertical, with 8 functions such as urban market, regional competition, second - hand housing dynamics, competitor analysis, opening monitoring, marketing public opinion, and policy interpretation. It can compress the professional real - estate report that originally required a team, 15 days, and tens of thousands of yuan in cost to about 10 minutes, reducing the cost by about 80%.
At the press conference on May 29th, when introducing these 8 functions, Zhou Xin, who has more than 30 years of real - estate industry experience, sighed, "I've been in the real - estate industry for 30 years. From the first day I entered the industry, I've hoped to have these 8 functions."
"From the first project, I've hoped to know the monthly and quarterly changes in the real - estate market of this city. I want to know the changes in the second - hand housing market of this city. I want to know the situation of my competitors. I want to know the situation of my region. I want to know the opening situation of the whole city, especially in my region. I also want to know the monthly public opinion in the whole city and its relationship with me, what kind of marketing each real - estate project in the whole city is doing, and what impacts the policies have on me." Zhou Xin said.
Now, many of the "I hopes" and "I needs" of real - estate people 30 years ago have been delivered to the new generation of real - estate people in the form of AI 30 years later.
This is not an AI created by an Internet giant to "seize the traffic entrance," but an AI truly created by a real - estate company according to industry needs, aiming to lower the industrial threshold for real - estate people, improve business efficiency, and serve "insider" users.
CRIC told 36Kr that in the future, CRIC Deep Intelligence Connect will further launch professional AI Agents for various scenarios such as second - hand housing agents, financial institutions, and property services, connect the upstream and downstream of the real - estate industry, cover more professional application scenarios, and meet the needs of more professional real - estate people.
This is why CRIC Deep Intelligence Connect is called "the first AI Agent in the real economy."
Actually, the real economy has embraced AI for a long time. Take the real - estate industry as an example. Many large real - estate developers have already used AI in internal land - value prediction, investment and financing management, housing - type design, supply - chain and construction management, etc. - but only within the company. After all, the company's own industrial data is the real lifeblood of the enterprise, and it doesn't matter much which large AI model, cloud computing, or CRM/ERP integration is used.
Similarly, in fields such as mining, metallurgy, transportation, and chemical industry, large enterprises tend to cooperate with large - model giants to customize privatized large AI models for the enterprise, which are also not open to the public.
And CRIC Deep Intelligence Connect is the first AI Agent that has a large amount of industry knowledge and know - how and is open to the industry.
To some extent, it represents