Zillow’s CTO says AI is reinventing every step of the home buying process

This month, Zillow officially celebrated the 20th anniversary of the real estate website’s debut. One executive who has served as a senior leader from the ground floor is David Beitel, who has served as CTO for Zillow’s entire existence.

“I really only had two jobs my whole career,” says Beitel. He initially joined Microsoft as a software design engineer in the 1990s and was slotted into a travel product division that would eventually become travel technology giant Expedia Group. It was there that he initially met two future co-founders of Zillow, Rich Barton and Lloyd Frank.

Throughout the past two decades, Beitel’s north star goal as CTO has been to infuse technology throughout all stages of the home buying process, making it easier for shoppers to search for a home, find an agent, schedule tours, make a successful offer, and complete a transaction. But he has also been keenly focused on internal use cases and says there’s not one part of the business where artificial intelligence isn’t being considered as a tool that can improve workflows.

Zillow’s engineers have been using AI coding assistant tools like Cursor and Claude for well over two years. Beitel says the secret sauce is that Zillow constantly tests these AI coding tools and models, making constant adjustments to ensure the team is using the best technology for the desired outcome. Zillow measures productivity, as well as the quality of the work these AI coding assistants produce. Overall, “the tools are getting very good,” says Beitel. “The models are improving every month.” 

Across the broader enterprise, every employee has access to OpenAI’s ChatGPT. Zillow also works closely with enterprise search and AI applications startup Glean, which hit $200 million in annual recurring revenue in December. Glean has made it easier for employees across Zillow to explore generative AI in a protected environment. IT has set up the proper guardrails to make sure workers can only tap data that they have permission to access. Using Glean, Zillow’s employees have already created more than 4,600 AI agents.

Beitel says employees explore AI at three distinct stages. The entry level is using AI to find information, then automation is one level higher. On the top floor, the most complex cases involve drastic workflow changes that would speed up the development of a new product or uncover a new pricing strategy or competitive analysis research that hadn’t yet been explored using AI. Now a few years into the generative AI journey, Beitel says Zillow’s workforce is expected to be at that third stage of adoption.

“Don’t just think of AI as an afterthought,” says Beitel. “But how do I use AI upfront to make the tasks I’m doing faster, more efficient, more effective, and more easily accomplished?”

And while Zillow’s business has been booming, with revenue totaling $2.6 billion in 2025, up 16% year over year, the broader housing market has been quite choppy. Last year, sales of previously owned homes hit their lowest level since 1995, as home buyers face mortgage rates that are double pandemic-era levels, high prices, and low inventory.

“If you’re a new home buyer, it’s daunting, and it’s a very important purchase—typically one of the biggest ones you might make in your life,” says Beitel. “So it’s natural that it would be fraught with some challenges.”

A key part of Zillow’s external AI story has been the company’s Zestimate feature, which uses AI and machine learning to analyze public records, market trends, and other data points to estimate the value of a home. Other technology investments have included leveraging computer vision models to create 3D walkthroughs of a home listing, AI-enabled virtual staging to help buyers envision what a home may look like with various design styles, and a newer feature called SkyTour, the latter uses drones to capture exterior images and give a potential buyer a fuller picture of what the neighborhood looks like. 

Beitel says LLMs are also making it easier for Zillow shoppers to ask natural language questions, rather than toggle through pre-set search criteria. One feature is called AI Assist, a conversational assistant that can answer renter questions including “Are utilities included in the rental?” or “Is there on-site parking?” 

Last year, the company also debuted functionality within ChatGPT. A ChatGPT user can write, “ZIllow, show me apartments in the West Village, New York City, for under $5,000 per month.” The AI chatbot will then pull various listings and link to Zillow’s website or app for additional information. Beitel says Zillow and OpenAI were able to ship the feature in about six weeks. 

AI is also being used for simple, mundane tasks like scheduling property visits. And in 2023, Zillow paid $400 million (plus $100 million in a potential cash earnout) for Follow Up Boss, a customer relationship management software system that uses AI to summarize calls and can suggest a to-do list for agents and personalized messages that can be sent to a potential client.

In the future, Zillow would like to infuse more AI into the more complex, tail-end stages of buying a home. Beitel says this work can be tedious and take a lot of time, as mortgage documents are shared between clients, buying and selling agents, loan officers, home inspectors, and more. AI could make this entire experience a bit smoother, Beitel says.

“We haven’t built it all yet,” he concedes. “We started at the top many years ago and we’re going deeper into the transaction.”

John Kell

Send thoughts or suggestions to CIO Intelligence here.

This story was originally featured on Fortune.com