Editor's note: This article is based on interviews with REALTOR® members who have articulated their personal preferences for AI tools based on their brokerage needs. NAR does not endorse any one AI tool or product. To learn more about which technology tools NAR partners with through the REACH program, visit https://www.nar.realtor/technology/reach
With AI products launching daily, deciding which to use at the brokerage level can be tricky. Rather than feeling pressure to invest in a handful, consider starting with a large language model (LLM), like ChatGPT by OpenAI, Gemini by Google or Claude by Anthropic. REALTOR® News talked with three top brokers who cited Claude as their “desert island” AI tool. Here’s what to know about Anthropic’s game-changing ecosystem.
First, What Exactly Is Claude?
Claude has evolved beyond a mere chatbot to an autonomous agent that can serve many functions. The technology is available as these products:
- Claude.ai: This consumer web, mobile and desktop product features a chatbot-style interface. The AI assistant comes in several plan tiers, ranging from Free to Enterprise.
- Claude Cowork: This collaborative AI assistant can autonomously plan and execute processes across a user’s desktop, files and applications.
- Claude Code: This agentic coding assistant for developers supports a protocol for connecting with external tools and data sources. (For more serious developers, Claude also offers an API and Developer Platform.)
- Claude Design: This past April, Anthropic launched this product for creating visual work, like designs and slides. A research preview is available for Claude Pro, Max, Team and Enterprise subscribers.
Across the Claude ecosystem, users choose from three models, each with different capabilities and pricing:
- Haiku is the lightest, fastest and least expensive. Haiku 4.5 is the latest version.
- Opus is the most powerful and pricey. Opus 5 is the latest version.
- Sonnet is the intermediate choice. Sonnet 5 is the latest version.
How Are Brokerages Using Claude?
Some brokerage leaders are using the technology to automate time-consuming processes and build custom tools. Others are finding that Claude helps them get a clearer view of what’s happening across their businesses. The level of adoption varies, but the examples below show how brokers are finding practical ways to use Claude to solve specific business problems and improve efficiency.
As with any AI tool, though, human oversight is essential. AI agents can take on much of the legwork involved in these processes, but brokerage leaders and agents still need to review their output, verify information and maintain control over final decisions.
The Ben Moss Group, Affiliated With Compass
Camila Castillo, luxury real estate specialist and AI expert at the Ben Moss Group in Coral Gables, Fla., hosts a weekly call to help her colleagues master Claude Cowork. Farming is one task she’s recommended they offload to the AI agent.
Recently, Castillo wanted to find prospects with high equity in their homes. First, she asked Cowork to generate a spreadsheet of properties bought seven or more years ago. Then she asked it to search the property appraisal website to ensure property addresses matched owners’ mailing addresses.
Normally, “that would take us weeks,” she says. “You can also do this with [PropertyKey’s] IMAPP, but sometimes it’s not accurate. Even if I were to pull addresses from IMAPP and the MLS, we would have to manually go into the property appraisal website to doublecheck if the mailing address is different than the property address.”
Castillo raves about Skills, a feature across the Claude ecosystem that enables users to create and save workflows. Castillo used to spend 45 minutes or more on each MLS property description, she said. Now, Cowork automatically handles this task. All she needs to do is input the address, executed listing agreement and last MLS tax roll for each new property she lists.
Raleigh Realty
Ryan Fitzgerald, owner of Raleigh Realty in North Carolina, was introduced to AI through ChatGPT. But once he discovered Claude, he was hooked. “Anybody using Claude is not going back to ChatGPT; Claude is so much better,” he says.
Fitzgerald relies on the entire Claude ecosystem, estimating it saves him 20 hours per week. “Claude can solve pretty much any problem in under 60 seconds,” he says.
In “under an hour,” Fitzgerald built a proprietary internal dashboard and reporting tool for his brokerage. “It runs the brokerage and enables agents to quickly and easily track their deals,” he says. “My agent asked me about reporting. So, I just went to Claude Cowork. I said, ‘Here is the problem—develop a solution.’ And the next thing I know, there’s an entire reporting dashboard built for me. I sent it to the entire team.”
Fitzgerald also used Claude Code to create AgentLoft, an AI-powered lead-generation platform for real estate professionals. Fitzgerald deployed the product within his company and, as of August 2025, has been selling it to other brokerages.
Bramlett Partners
Texas-based Bramlett Partners has been building its own code for years, said broker-owner Eric Bramlett. About a year and a half ago, Bramlett started using the developer tool, Replit, to “vibe code,” or prompt AI to write, test and refine code for him.
He switched to Claude Opus after version 4.6 launched earlier this year. “There has been an arms race between different AI companies and models leapfrogging each other,” Bramlett says. “But Claude’s Opus 4.6 was a massive leap forward for agentic coding. All of a sudden, it allowed non-coders to push out code that’s well-written and can scale if you know what you’re doing.”
With Opus 4.6 and five developers, Bramlett created his “dream” brokerage management platform in about 100 hours or three months. Before Opus 4.6, the project would have cost $2 million, he estimated.
“Something like this [platform] did not exist one year ago,” he says. “It would have been so expensive; nobody would have built it. But as AI progressed, it became possible and affordable. Because of agentic coding, we can build at 10 to 15 times the speed of traditional development.”
The platform Bramlett built features one dashboard with 17 AI agents, each devoted to a specific task, like monitoring emails or phone calls or checking for compliance.
Along with improving his agents’ efficiency, the platform helps Bramlett run his brokerage. “It knows everything and gives you full 360-degree visibility into your business,” he says. “We’re able to make decisions quickly and with a lot of confidence because we have every piece of data piping into one location. It helps me sleep well at night since I have an agent on top of it and can ask questions."
The AI agents surface information, identify potential issues and help Bramlett understand what needs his attention, but they don’t replace human judgment. Bramlett is responsible for reviewing that information and making the final decisions about how to act on it.
He continues, “I asked it yesterday how the business was going and if there was anything I needed to worry about. It ran through every piece of the business and said, ‘Your email deliverables look great, your revenue looks great, don’t mess with it. But you have one process that’s broken and need to fix.’”
Now that his platform is complete, Bramlett is using Opus 4.7 to build a CRM—the only remaining component in his brokerage’s tech stack that is not proprietary, he said.
The AI Tools Your Agents Are Using
Even if your brokerage isn’t using AI to build systems or scale, your agents are more than likely using one of the many AI tools available to them.
According to NAR’s 2025 REALTORS® Technology Survey, the AI tools residential real estate agents are relying on the most (including the percentages using them) are:
- Writing tools (78%)
- Chatbots and virtual assistants (47%)
- Image editing and virtual staging (39%)
- Market analysis and automated CMA tools (39%)
- Predictive consumer analytics (41%)
- AI-enhanced drone photography and video (52%)
It is imperative that brokerages have an AI policy in place to reduce risk and protect clients, including clear expectations for when human review and approval are required. AI can streamline tasks and surface information, but brokerage leaders and agents remain responsible for the decisions they make and the work they deliver. If you don’t have a policy in place or want to make sure yours includes everything it should, this customizable AI policy template is a good place to start.
