A growing group of real estate agents are reporting that AI is improving their productivity. It just comes from outside the brokerage firm’s technology stack.
Real estate agents are finally starting to see productivity gains from artificial intelligence, but those gains aren’t coming from in-house technology.
And these tools are being used by a growing percentage of agents to, in their words, “significantly” improve their work performance.
The latest results from the Intel Index study show significant increases in AI adoption across categories, particularly in image editing and market data analysis.
However, they also suggest that recent developments in AI tools provided by brokerages have so far failed to replace or even rise to parity with external tools favored by agents such as ChatGPT, Claude, and Gemini.
Intel set out to better understand how agents actually use AI. In our April study, Intel looked at questions such as whether agents still primarily use chat interfaces or are starting to experiment with more integrated and autonomous tools.
Read the first findings in this first article in our series on AI in the securities industry.
Agents are leveraging AI tools, but who are they?
In previous research, the majority of agents using AI only generated text for listing descriptions and social media marketing.
That’s starting to change. And as a result, agents self-report greater productivity than they previously reported.
The majority of agents responding to an Intel survey (70% in April) reported seeing at least some productivity improvement from AI. This is up from 62% the last time Intel asked this question in August.
And for a growing number of agents, increased productivity is having a huge impact.
The percentage of agent respondents who told Intel that AI tools have “significantly improved” their work productivity increased from 11 percent in December 2024 to 15 percent last August, and reached 28 percent in April.
Despite these benefits, real estate agents seem to maintain a healthy degree of skepticism regarding the accuracy of the output of these tools.
However, few people consider accuracy to be a significant issue that prevents the tool from being used.
When Intel first surveyed agents on AI use in the final weeks of 2024, about one in six respondents said they believed text generated by AI was fundamentally untrustworthy. Since then, that share has fallen by nearly half, to 1 in 11 people.
Instead, the majority of agents surveyed said that AI output is generally good enough to be useful, but requires some level of fact-checking.
Thirty-six percent of agent respondents said they believe AI results are usually accurate, but that “rarely” factual errors occur. 49% said they expected AI to make factual errors “frequently” but not enough to deter them from using it.
All of these stock prices in April were almost unchanged from the August survey.
For the first time, Intel also began tracking agent adoption by agents such as Claude Code, Claude Cowork, OpenAI’s Codex, open source OpenClaw, or another agent AI platform offered by a brokerage firm.
These environments are becoming an increasingly popular way for users to interact with AI tools, especially in areas such as software development.
Instead of traditional chat interfaces where users manually upload documents or other files and chat with an AI model to get answers or generate simple output, these agent tools give AI models direct access to files on their computer or in a software suite. This allows the model to perform multi-step actions on behalf of the user.
However, Intel found that very few real estate agents are leveraging agent AI in their operations.
In April, 75% of agent respondents said they primarily use conversational chat interfaces when interacting with AI tools. In contrast, only 9% said they primarily use agent or integrated environments such as Claude Code or Cowork.
Despite the introduction of a number of AI tools for brokerage that have gained attention in recent months, real estate agents seem to be primarily using tools from OpenAI, Anthropic, and Google rather than in-house tools in their daily work.
71% of April agent respondents said they use their preferred AI tools more often than those provided by their brokerage firm. It’s actually increased slightly since August. By comparison, 10% said they use their brokerage firm’s AI tools most often, and 19% said the same percentage.
In particular, agents who primarily rely on their brokerage firm’s AI tools rather than their own external tools are less likely to say they’ve seen the big productivity gains that other agents are reporting.
Among agent respondents who primarily rely on brokerage AI tools, only 17% said they had experienced “significant” productivity gains from AI in their work. Almost twice as many agents (33%) who primarily use their own external tools say their productivity has increased significantly.
These will be key areas to focus on in the coming months as brokerages pour resources into developing internal tools and platforms that leverage these models.
Intel’s research also revealed the top ways agency and brokerage leaders are using AI in their business.
In next week’s report, we’ll examine the specific ways in which the average agent is leveraging these tools more and more over time, and how a small number of AI power users are realizing the biggest productivity gains available today.
Methodology note: This month’s Inman Intel Index survey was conducted from April 22nd to May 5th and received 435 responses. The entire Inman reader community was invited to participate, and a rotating selection of randomly selected community members were encouraged to participate via email. Users answered a series of questions about their self-proclaimed niche in the real estate industry, including real estate agents, brokerage leaders, financiers, and proptech entrepreneurs. Results reflect the views of our passionate Inman community, but do not necessarily align with the views of the broader real estate industry. This survey is conducted monthly.
Email Daniel Huston
