Don’t miss: Inside AI Episode 6: Justin Lundy, founder and CEO of Lundy, explains what happens when AI becomes the front door to the technology agents use every day
For years, real estate professionals have complained about having too many technology tools. Your MLS gives you one set. Your brokerage gives you another. Your coach tells you that you need three more. Then you add your own CRM, transaction management platform, forms software, email, calendar, CMA system and whatever else makes your business work.
Justin Lundy knows that problem firsthand because he lived it as a real estate agent. At one point, he had about 15 different technologies. The problem wasn’t necessarily that the tools were bad. It was that he became what he calls “human middleware,” constantly moving information from one system to another because none of them really worked together.
Now Lundy is trying to remove the human from the middleware job without removing the human from the decision-making. That distinction may become one of the most important things we need to understand about agentic AI.
In Episode 6 of Inside AI with The REAL AI Guy, I sat down with Justin Lundy, founder and CEO of Lundy, to talk about Nora, the company’s new AI assistant for real estate. But the more interesting story may be how Justin got here, because his journey explains a lot about the way he thinks about technology, accessibility, voice, risk and why humans still need to remain firmly in control.
The problem started with too many text messages
Justin didn’t start in AI. He started in real estate. He was part of a small San Diego team handling between 30 and 50 listings a month. That kind of volume created some unexpected bottlenecks. Agents walking through the listings with buyers would constantly text questions that were already answered somewhere in the MLS or marketing material.
What kind of countertops are these? What is the flooring? When was this remodeled? Multiply those questions across dozens of listings and hundreds of showings, and suddenly answering texts becomes part of somebody’s full-time job.
Justin had been learning to code and stumbled across the idea of building Alexa skills. This was long before ChatGPT and today’s large language models. Alexa wasn’t thinking. You had to anticipate what someone might ask and program the answer.
So Justin built a real estate Alexa skill and started placing Alexa devices inside his listings. It worked. The number of questions coming into the team dropped, and buyers could ask Alexa directly about things they wanted to know about the property.
The idea even became part of his team’s listing presentation. It was a huge competitive differentiator: Sellers could use it to communicate details about their home that would never fit in an MLS field.
That could have been the end of the story. Instead, something much more important happened.
What his mother-in-law taught him about technology
Around the same time, Justin’s mother-in-law began losing her eyesight from retinitis pigmentosa, a condition that progressively reduces peripheral vision. That experience changed his perspective.
Justin began helping her understand the resources available to someone losing their vision. Should she use a cane? A guide dog? What tools were available? Through her, he also began meeting people in the blind community and learning about everyday experiences most sighted people never have to think about.
He talked about going to dinners, spending time with people who were born blind or had lost their sight. Justin began to understand, in his words, “what the world didn’t include for them.”
That line stuck with me. Technology people constantly talk about building something new. Justin was learning something different: Sometimes innovation comes from finally noticing who the old technology left out.
Then the connection became obvious. “Of course, the blind would love to search by voice,” Justin said in the interview. That realization led Lundy to create Finding Homes, a voice-based property search platform. The company partnered with the National Federation of the Blind, which organized focus groups of blind users around the country to test the product, break it and teach Lundy what an accessible real estate search experience actually needed to be.
That may be one of the reasons Lundy’s work today is so voice-focused. For Justin, voice was never a gimmick. It solved a real human problem first.
Then GPT changed everything
Lundy was already working on conversational real estate search when Justin saw some early GPT examples posted on Twitter by technology innovator Marc Andreessen. At first, he kept scrolling by. Then he saw another example. And another. Eventually, it motivated him to check out ChatGPT.
For a company that had spent years manually programming the language behind voice search, the implication was immediate. Suddenly, the system didn’t need every possible noun, verb and phrasing manually defined in dozens of spreadsheets.
Justin sent the technology to his CTO. His CTO initially pushed back. He explained that technology goes through hype cycles all the time, and things rarely change as much as people think they will.
That was on a Friday. By Monday, Justin said his CTO came back with a very different answer: “I was wrong. Everything’s about to change.”
Then came the kicker. “We can throw away about 70% of our code if you want.”
Justin’s response? “Let’s do it.”
The first time they connected GPT to their voice system, Justin said it gave him chills. Years of carefully constructed technology suddenly looked old almost overnight.
That’s a feeling a lot of us in AI have experienced over the last few years. You build something that feels advanced, and then the underlying model improves and changes, which moves the goal posts, again.
Why his AI assistant is named Nora
Before getting into what Nora does, I asked Justin to share where the name came from. It is now one of my favorite stories for naming an AI product.
Justin and his co-founder both had daughters around the same time. They never discussed names. Justin said the day his co-founder’s daughter was born, he received the introduction: “Meet Nora.”
His reaction was immediate. “You got to be kidding me! That’s our name.”
Their daughters are about a month apart in age. Both are now six. Both are named Nora. As Justin put it, “We got a lot of Noras running around.”
When it came time to name the company’s AI assistant, the answer was sitting right in front of them. “Should we just call it Nora?” They did. And Justin said in the interview that his daughter loves it.
There’s something fitting about that story because Lundy isn’t trying to create some faceless robot overlord. The goal is much more practical. Make the technology feel like something you can simply talk to.
Real estate doesn’t need 15 AI assistants
Justin thinks nearly every real estate professional will eventually have a preferred AI assistant. Maybe that’s Nora. Maybe it’s ChatGPT, Claude, Gemini or something we haven’t heard of yet.
His bigger prediction is that the assistant becomes the front end while today’s real estate technology becomes the infrastructure underneath it. CRMs don’t disappear. MLSs don’t disappear. Transaction management systems don’t disappear. You just stop having to live inside every interface.
That’s why Justin describes Nora as an “orchestrator.”
That is a powerful idea because real estate has been repeating the same technology mistake for decades: building silo after silo. Now we are doing it again with AI.
Every platform is creating its own AI assistant. The CRM has one. Transaction management has one. The MLS may have one. Your brokerage has one. Justin joked that soon agents may need AI just to remember the names of all their AI assistants.
The smarter solution is not necessarily another silo. It is an AI layer that can work across them.
Talk-to-text is not the same as voice
This is another area where Justin’s years of voice experience matter. A lot of companies are adding microphones to software and calling the result voice AI.
Justin sees a big difference. “Voice doesn’t mean slap a microphone on your tool and let it transcribe what I’m saying,” he said in the interview.
His phrase for the difference is excellent: “It’s not talk to text. It’s talk to understanding.”
That is exactly where voice interfaces are headed. The goal should not be to dictate instructions into a box. The goal is to communicate naturally with technology and have the system understand your intent.
For a real estate agent, that could mean leaving a showing and telling your AI assistant to prepare a CMA on the drive home. It could mean asking it to monitor for an offer to arrive by email. It could mean asking it to watch for a price change on a property or collect your calendar appointments, important emails and provide yesterday’s meeting summaries before your day begins.
That is very different from asking ChatGPT to help you write an email. This is AI moving from generating to doing. And that is where I start getting nervous.
The stakes change when AI can take action
I’ve been saying this repeatedly: Agentic AI scares the bejeebies out of me. Not because I think we should stop using it. The opportunity is enormous.
But when an AI system can access your email, calendar, MLS, forms and transaction systems, a hallucination is no longer just a bad paragraph. It introduces a massive risk.
That means the guardrails must be much stronger. This was one of the most important parts of my conversation with Justin because Lundy has deliberately designed friction into Nora.
Before Nora can do something permanent, such as send or delete something or move forward with writing an offer, the human must confirm it. Not verbally. You cannot simply tell Nora, “Yeah, that looks good. Send it.”
Justin said those actions are placed behind a hard-coded confirmation button. The user must physically approve the action.
That is my HIM – or “human in the middle” principle I teach in my AI classes for agents.
AI earns our trust incredibly quickly, sometimes faster than it deserves. The better these systems become, the easier it is to stop checking them. A confirmation step forces you to look one more time.
The future may arrive faster than agents expect
Near the end of our conversation, I asked Justin what the next 18 months might look like. His answer surprised me because he doesn’t expect the typical technology adoption curve.
He thinks AI assistants could move much faster. “In the next 18 months, you know what we are doing now will seem old,” he said.
Then he described what he expects when agents realize they may no longer have to log into every piece of technology separately. “I think it’s going to be a tidal wave.”
He may be right.
Glad I work for the WAV Group (terrible pun intended). Enjoy more from Justin by watching Inside AI with The REAL AI Guy: Episode 6 with Justin Lundy here: https://youtu.be/AmRxx0tIUFI
# # #
Kevin Hawkins, a partner with WAV Group, is editor and co-founder of REAL AI, real estate’s No. 1 AI newsletter, its weekly podcast and the Inside AI with The REAL AI Guy video interview series. He has written more about AI than anyone in real estate. Hawkins is the Amazon bestselling author of The REAL AI Guide for Real Estate Agents and is known throughout the industry as The REAL AI Guy, teaching thousands of real estate agents how to use AI in the best and safest ways.
The post Inside AI – Episode 6 with Justin Lundy – The AI inflection point: When AI stops answering and starts doing appeared first on WAV Group Consulting.
Leave a Reply