Why Doesn’t AI Work for Us?
Before we get into it, here’s a bit about me:
I’ve spent the last 15 years building software products, leading ops, and helping companies scale. In 2009, I co-founded a SaaS startup in the logistics space, had a successful exit (and picked up a few grey hairs along the way), then launched a consulting practice that eventually led me into real estate. Now, I’m building AI tools designed for how this industry actually works, not how outsiders think it works. If you’ve ever felt like the tech you’re handed doesn’t match the chaos you’re dealing with, I get it. I’ve lived it. Solving problems with software is my thing, and I’m glad to have a little corner of the internet to share what I’ve learned and what I’m still learning as AI continues to shift the ground beneath us.
This is why I created What the FAQ?! It’s not just another AI opinion column. It’s a weekly breakdown of how AI can actually work in real estate, one practical question at a time.
Let’s start with the big one:
why doesn’t AI work for us?
If you’ve tried AI and walked away unimpressed, you’re not alone. Maybe it was Zoom AI Companion summarizing your meeting and missing every important detail. Or Microsoft Copilot in Teams confidently suggesting answers that made no sense in context. These tools often sound promising, but they weren’t built for real estate and definitely not for the complexity your team handles every day.
Most real estate teams have been given generic AI that doesn’t accurately reflect how their business operates and that’s the core of the problem. AI only works when it knows what it’s talking about. Unless it has been trained on your docs, your policies, your onboarding decks, and your daily fire drills, it’s not going to be helpful. It’ll give you guesses. It’ll get things wrong, and it’ll leave your team cleaning up the mess.
Most AI tools show up pre-trained on general web data or open-source content. That might work for a restaurant or a SaaS company, but in real estate? You’ve got layers of rules, local policies, acronyms, legacy workflows, and a whole lot of internal knowledge that doesn’t live in public forums. That’s why agents end up frustrated. They ask a question, get a half-right answer, and then call your team anyway.
The fix is surprisingly simple. Feed the AI your actual knowledge. Drop in your PDFs, your Word docs, your help center articles, your training guides, your policies - whatever information you want made available. Let it learn from your real-world materials, then test it. Ask the questions you hear every day. If it misses something, update the source. Over time, the tool doesn’t just get better, it fills the gaps in your knowledge base and becomes your company expert.
How real estate teams are making AI work
The organizations achieving real results with AI are those doing a few key things right. First, they train the AI on internal documentation. That includes policies, FAQs, benefits, scripts, and other material that reflect their actual operations, not just best guesses. Next, they test the system using real agent questions. If it fumbles something, they use that as a prompt to retrain or improve the underlying knowledge base.
Your AI should get better over time.
You’re not aiming for perfection on day one, and I think there’s something kind of beautiful about that. It means your members get to play a role in shaping something that’s built for them. That’s what good service looks like.
So yes, organizations achieving results are doing it by keeping their knowledge base updated, just like they would a staff playbook or help center, and they track results instead of assuming everything is working behind the scenes. This isn’t just another tool to bolt on. It’s a smarter extension of your team. And it should be managed, measured, and maintained like one.
What are they measuring?
The teams getting value from AI aren’t winging it. They’re watching specific metrics to understand what’s working. Support volume reduction is an important one. Some MLSs are seeing 40 to 60 percent fewer inbound support requests after launching Voiceflip’s AI assistant, Ardi. That kind of drop frees up time and headspace for staff. Time saved per inquiry might seem like a small metric, but it adds up quickly. Saving even two or three minutes per interaction across hundreds of questions each week means hours recovered every month. Some Voiceflip clients are seeing improvement in their internal SLA performance. With fewer tickets to triage, they’re meeting response time goals faster and with less stress.
And yes, agent sentiment is trackable. Some organizations are using one-click feedback surveys after each interaction, and satisfaction scores increase when the answer is both fast and accurate.
A few big-picture stats
The AI in the real estate market is projected to grow from $222 billion in 2024 to $303 billion in 2025.
Currently, only 36% of real estate organizations are utilizing AI. By 2030, that number is expected to reach 90%.
Businesses already using AI report a 7.3% increase in productivity, along with measurable improvements in customer interaction and operational efficiency.
Sources: Brainvire The Business Research Company
What This Means For You
AI isn't here to replace your team. It’s here to lighten the load and enable your team to better perform. Done correctly, it cuts down the noise, delivers fast answers, and frees your staff up to focus on the work that matters, like building relationships, improving member experience, and taking on projects that drive real value. And the best part? You don’t have to overhaul your entire system to make it work. Start small. Train it on what you already know. Measure the results. Iterate from there.
That’s how you make AI useful and worth the investment.
Up Next Week:
Why Does My Staff Hate Using AI?
We’ll break down the most common rollout mistakes and how to fix them without resorting to donut bribery. Though if you want to bring donuts, no judgment.