How much time does the average commercial real estate broker save by using an artificial intelligence assistant? The answer from ChatGPT itself: 10 to 20 hours per week, depending on how fully the broker integrates AI tools into their workflow.

Even as AI has become more entrenched in background workflow processes, real estate brokers are navigating a growing universe of large language model tools like ChatGPT, Claude, Gemini, Perplexity and OpenAI o3. They’re also wrapping their minds around practical use cases and best practices to create efficiency without getting tripped up by the “hallucinations” or mistakes that these virtual assistants are still apt to make.

“Everyone uses AI day-to-day whether they realize it or not because so many things are tied into the back of any given workflow, whether I’m making a restaurant reservation, organizing my calendar or drafting emails,” said Mitchell Hernandez, a partner and co-founder of Beta, a California based commercial real estate advisory firm. Common tasks for which brokers rely on AI assistants these days range from abstracting leases and creating marketing content to analyzing markets and developing research reports.

Talking about how brokers are using AI is really talking about the tech stack brokers are using because AI has become an integral part of a variety of different tools and solutions. “We’re getting more used to AI, and I think that’s really telling,” said Hernandez. “I’ve lost count of the various tools that I’m using for day-to-day task management and project management.”

Practical Use Cases: Putting AI to Work

AI is removing a lot of the repetitive tasks and problems that need to be solved. “It’s freeing up more of a broker’s time to do the most important part of their job, which is the consultative and advisory piece,” said Laura Barr, Americas retail leader for CBRE Advisory and Transaction Services.

CBRE is leaning into agentic digital assistants to help its brokers improve their workflows. Its proprietary internal AI platform, Ellis AI, aims to boost brokers’ productivity by assisting with manual tasks in research, data analysis, contract abstracts and other areas. “For anything serving a client, AI can be helpful, but it is helpful behind the scenes,” said Barr.

For example, AI can summarize, analyze or search large documents or blocks of information quickly. She noted that an AI assistant is a good tool for a first draft or a first cut at analyzing a particular data set. In some cases, an AI model can complete as much as 80% to 85% of the preliminary work on a particular task. However, such projects need human touch and human oversight to take them the rest of the way, she said.

“At SRS, AI isn’t a buzzword; it’s quickly becoming a strategic pillar,” said SRS Real Estate Partners president Garrett Colburn. “We’re embedding artificial intelligence into the core of our operations to strengthen our real estate advisory and transaction capabilities.” The company is working with its brokers and staff to co-develop tools that streamline high-effort, time-consuming tasks like data discovery, lease abstraction and void and market intel analysis. “Traditional workflows often involve multiple steps, manual handoffs, emails and time delays between each phase,” said Colburn. “Our AI tools are changing that, cutting processes that used to take hours or days down to just seconds. It’s not just about efficiency; it’s about transforming how we work so we can focus more on value, strategy and client success.”

What Everyone Wants: More Time

Brokers are embracing AI tools for their efficiency and time savings. “Real estate ultimately is a game of speed,” said Tanner Olson, managing partner at Legend Partners, which is part of the X Team Retail Advisors network. “Time kills all deals, so just the ability to expedite deals … is always going to be beneficial.”

Olson uses AI chatbots on tasks like drafting lease extensions. For example, a broker can feed a lease document into an AI tool and can give the tool a prompt like: “Extend this lease for five years with 3% rent bumps.” The tool can read the entire lease, absorb the pertinent information and create a simple addendum that outlines an extension.

He also uses AI to track down property ownership information. Manually searching for those nuggets had been a tedious, time-consuming process. Now, AI tools can scan webpages to identify properties’ true owners and their contact information. “It’s a much more seamless and instantaneous process,” he said.

Avoiding Hallucinations

Brokers need to navigate around AI pitfalls, such as exposing private client data and generating “hallucinations” or otherwise false or misleading information. Ways to avoid errors are to evaluate the sources of the information the AI spits out and to use different AI tools to cross-reference findings.

Hernandez recently used AI tools to form a location strategy for a retailer rolling out to new markets. He had to drill down into various markets’ zoning requirements and conditional use permit use cases. Such analysis typically would have involved 30 to 60 hours to make phone calls, dig through records and identify points of contact for each municipality. With the help of AI tools, however, Hernandez built a fully cited preliminary research report in about 10 minutes. He then went through individual sections and cross-referenced results from multiple AI tools to make sure all was cited correctly. That review did reveal some mistakes. For example, the zoning information AI produced for one city wasn’t accurate; it was actually zoning information for another city. However, the fact that AI tools can do a majority of the work quickly is a “phenomenal” time-saver, said Hernandez.

While brokers sift through the various AI tools and models, brokerage firms are figuring out policies and best practices to generate results without costly mistakes. For example, SRS is balancing cost with quality and security. “While some free tools exist, they don’t protect client data,” noted Colburn. SRS is building proprietary solutions that integrate with its existing data environment, are compliant and secure and can scale. “Investing in the right tools now will pay dividends in the future by making our teams more efficient and our offerings more competitive,” he said.

Jumping on the Learning Curve

“This still feels very much like the Wild West,” said Hernandez. AI is moving quickly, and keeping up requires ongoing learning. ChatGPT first hit the market in fall 2023, for example, but it wasn’t until fall 2024 that a standard way for large language models to integrate with other pieces of the puzzle emerged, noted Hernandez. Model Context Protocol’s user guide explains itself via analogy: “Just as USB-C provides a standardized way to connect your devices to various peripherals and accessories, MCP provides a standardized way to connect AI models to different data sources and tools.” MCP connections help get data out of silos and help AI assistants produce better, more relevant responses.

Brokers also need to make time to practice with AI assistants and to understand how the tools work. CBRE has regular training for those brokers who want to learn how to use AI tools better. For example, on group video calls, CBRE brokers share case studies for the benefit of their peers.

But “no matter how much we train, if you don’t actually spend time with these tools, you will not truly learn and understand how to use them,” said Barr. For example, inputting the same question into different AI models delivers a better sense of the different answers one can get and which tool might work better for different uses.

CBRE is talking a lot with its brokers about inputting good questions or prompts in order to get good data or answers. “The better the question and query, the more interesting the response that comes back,” said Barr. Some people are getting really good at writing complex queries to extract great answers, she said.

AI can be overwhelming when one hears about all it can do. Thus, getting started can seem daunting, said Olson. He advises starting with one task or problem and figuring out how AI can help. “Really get comfortable with using AI before moving on to the next thing,” he said.

MORE FROM C+CT: 4 Quick Tips for Using AI Effectively