A new type of AI that could impact real estate is starting to gain traction among tech geeks and innovators. It is an autonomous AI “agent” that can manage complex workflows across multiple software platforms.
One of the most talked about systems among developers is OpenClaw (formerly Moltbot, Clawdbot, and Molty). It’s an open-source platform that allows users to build AI assistants that can perform tasks across apps, websites, and databases.
OpenClaw gained huge popularity in late January, following the Moltbook project, a Reddit-like “social media” site designed specifically for AI agents. Last month, OpenClaw developer Peter Steinberger announced he was joining OpenAI.
Although the technology is still largely experimental, Canadian real estate broker Daniel Foch said he has been testing OpenClaw, which specializes in real estate workflows, for a month and believes the industry could greatly benefit from it.
Foch is the Chief Real Estate Officer of Valery.ca and host of Canada’s Top Ranked Real Estate Podcast. He is also the co-founder of The Habistat, the data science platform behind TRREB and Proptx. The platform contributes to greater transparency in the housing market by providing real-time data and analytics that inform decision-making across the industry.
“As far as I know, no one in the real estate industry has spent more time making OpenClaw work for real estate agents than I have,” Foch said in a recent YouTube video documenting his experiments with the platform. “I truly believe that real estate will benefit more than any other industry from technology like this.”
“On-site work using desk backend”
Unlike typical AI chatbots that simply generate text or answer questions, OpenClaw is designed to perform tasks autonomously. Or, as their website says, it’s “AI that actually does things.”
This system can connect to software tools such as email platforms, calendars, databases, and CRM systems. Once you configure your AI models, you can execute instructions across those systems. This feature can be especially useful for real estate agents whose jobs often involve juggling multiple software platforms while spending a lot of time outside the office.
“Real estate is a field job with a desk backend,” Foch said in the video. “Agents are in their cars going to showings, inspections, listings, etc. They’re not sitting in a CRM or email all day.” He said the disconnect between field work and administrative tasks is creating friction in the industry that AI automation could help alleviate.
AI takes over the boring tasks
Foch says his customized OpenClaw system now handles a wide range of tasks within the real estate business.
Among other things, he says, AI assistants can update contacts and follow-up tasks within a CRM, generate listing presentations and home appraisals, schedule and add showings to calendars, organize paperwork and transaction documents, and draft outreach campaigns to potential customers.
The system can also generate content for marketing and media channels, such as creating podcast summaries, newsletter posts, and YouTube scripts. In total, Foch estimates that AI assistants can now handle roughly 70 to 80 percent of the administrative tasks that previously took up 20 to 25 hours a week.
“Before this system was in place, we were spending a tremendous amount of time context-switching between tasks, such as formatting documents, updating systems, and managing outreach,” he said. Now, he says, AI manages most of these tasks and he and his team oversee the process.
“Screenless CRM management”
To make OpenClaw convenient for agents, Foch says they have built integrations with several popular real estate software platforms.
His system can use APIs to interact with CRM tools commonly used by agents, such as Follow Up Boss and kvCORE. We also created automations that allow AI to manage lead follow-up, launch nurturing campaigns, and update customer records.
As an example, he said, you can send a text message to the AI assistant telling it to move a contact into a nurture sequence and schedule a follow-up call. The AI then updates the CRM and triggers the appropriate automations to draft follow-up messages. He says this kind of “screenless CRM management” is especially valuable for agents who spend a lot of time driving between showings and meetings.
Why most agents don’t use OpenClaw
Despite OpenClaw’s potential, it is far from mainstream adoption among real estate professionals.
One reason is complexity. Setting up the system requires technical knowledge, including configuring integrations, building workflows, and defining detailed standard operating procedures to guide the AI’s behavior.
Foch said he spent a month documenting the process and creating an open-source “skill” that agents could use to reproduce the configuration. Still, he said most agents who review the material conclude that building the system themselves would require a significant amount of time and technical expertise.
“People keep telling me they just want the profits,” he says. “They don’t want to go through the whole process of building it.”
In response, Foch and several partners are developing commercial products that package OpenClaw-based automation into ready-to-use systems for real estate professionals. The product, called Homie, aims to provide agents with a preconfigured AI assistant that integrates with CRM, phone, and email systems.
The team plans to launch the platform with a limited number of users initially and refine onboarding and support. While it remains unclear whether systems like OpenClaw will be widely adopted in real estate, Foch believes the technology will revolutionize the way agents manage their businesses.
“The industry doesn’t necessarily need direction,” he said. “I need to pack it.”
‘Absolute nightmare’ for cybersecurity
There is also another issue: cybersecurity.
OpenClaw’s powerful capabilities do not mean that real estate agents need to rush into implementing OpenClaw. According to security researchers at Cisco Systems, this technology poses significant cybersecurity risks, especially if AI agents are given access to files, system commands, and third-party applications.
“From a feature perspective, OpenClaw is a breakthrough. It’s everything that developers of personal AI assistants have always wanted to achieve. From a security perspective, it’s an absolute nightmare,” Cisco’s Amy Chang and Vineeth Sai Narajala recently wrote.
Because OpenClaw can execute shell commands, read and write files, and run scripts directly on a user’s machine, granting high-level privileges to an AI agent could potentially allow it to perform harmful actions if misconfigured or if a user installs a malicious or compromised “skill.”
Researchers have also reported instances where OpenClaw exposed cleartext API keys and credentials. These credentials can be stolen by an attacker through prompt injection techniques or insecure endpoints.
The OpenClaw documentation also acknowledges this challenge. According to the project’s own guidance, “no completely secure setup exists.”
Wired reported last month that Meta and several other technology companies have internally restricted the use of OpenClaw due to security concerns. One Meta executive said he recently told his team not to install software on company laptops, warning that doing so could lead to termination.
The executive, speaking on condition of anonymity to discuss internal policies, said he believes the tool is unpredictable and could potentially pose a privacy risk if deployed within secure systems.
That doesn’t mean Meta isn’t interested in the technology. The tech giant recently acquired Moltbook, a “social network” for AI agents built on OpenClaw. A Meta spokesperson confirmed that Moltbook will join the company’s Meta SuperIntelligence Labs. Terms of the deal were not disclosed.
A preview of the AI agent era
For now, OpenClaw remains primarily the domain of developers and tech-savvy real estate professionals like Foch. For many agents, the cybersecurity risks outweigh the benefits. Instead, most people will take advantage of similar capabilities through packaged software platforms (such as emerging systems like Foch’s Homie or Breezy) rather than building AI agents themselves.
Still, the capabilities Foch demonstrated with OpenClaw are an early indication of what the future of “agent” AI in real estate could look like, with AI handling the tedious back-office tasks while real estate agents focus on the IRL side of the business.
Email Nick Pipitone
