Ed Neuhaus 
She had just finished writing the last check to her developer. Six months of work, a custom site with Elementor under the hood, and a property search bundled in for $150 a month. She was about to sign up for a second site, the one all her agent friends swore by for SEO, that would have added another $350 a month on top. So 90 days after launch, she rebuilt the whole thing herself with AI and saved $500 a month, every month, forever.
That’s Ariella Faitelson, an Atlanta agent with Vesta Consulting Group at Residential Realty. Her whole brand is education-first, and you can feel it the second you land on her site. The part that gets me, the part I keep coming back to, is that she had a perfectly fine website when she did this. Most agents in her spot would have lived with it. She didn’t.
The trap of a developer-built site
Lets talk about why this matters, because Ariella’s situation is not rare. It might actually be the default.
You hire a developer. You spend six months going back and forth on revisions. You finally launch. The site looks great. And then a week later you realize you want to change a headline on the homepage, and you cant. You email the developer. You wait. They invoice you for the change. The next time you want to add a neighborhood guide, same thing. The next time you want to swap a photo, same thing.
The site is yours but you dont own it. Not really. You own the rights to it. The developer owns the keys.
Elementor is the same trick wearing different pants. It’s a great tool for developers because it gives them a flexible canvas to build on. For agents, it’s a maze. The widgets layer on top of widgets, the styling overrides override other styling, and one wrong click breaks something three pages deep. So even though you technically have access, you dont touch it because you’ll break it.
Ariella’s site wasnt bad. It was actually pretty good. But she described it as feeling “heavy and clunky” and the part that mattered most, she couldnt change a single thing herself without picking up the phone. That’s the universal problem. Every agent I know with a developer-built site has the same complaint, even when they like how it looks.
The lightbulb moment
Ariella saw me posting on Facebook about the AI-built version of neuhausre.com and what it was producing in terms of traffic and leads. She read the comments. She asked questions. And in her own words, “I realized there was a way for me to rebuild it after looking through your comments that would give me more results and that I could be in control of.”
That last part is the part nobody talks about. Control.
When you built your business around someone else’s tool, whether that’s a developer or a $350 a month SEO platform, you are renting your reputation from them. They go down, you go down. They raise rates, you pay. They retire, you start over. None of that is hypothetical. I’ve watched agents lose entire content libraries because a vendor sunset a product. Seth Godin’s whole point in Linchpin is that the people who build their own platforms are the ones who outlast the gatekeepers. Ariella read that signal and acted on it.
How she actually did it
Here’s the part I want every agent reading this to understand. She didnt have AI write her website. She built a system that uses AI as a writer, with her in the editor’s chair.
The workflow looks like this. She built a voice profile, a structured document that captures how she actually talks, what she cares about, the words she uses, the words she doesnt. Then for any new piece of content, she brainstorms the angle herself, outlines the structure herself, and then has AI draft inside her voice. She proofs every piece. She rewrites where it sounds off.
That’s a system, and systems are what scale. She didnt approach AI as a shortcut, she approached it like a process. Outline, draft, revise, publish. Anyone can ask Claude to write a blog post. Almost nobody bothers to build the voice profile that makes the output sound like them.
It’s also why her site doesnt read like AI slop. The content is grounded, specific, written for the buyer in Decatur or Buckhead who’s actually trying to figure out where to live. That’s not what you get when you type “write me a blog post about Atlanta real estate” into a chatbot. That’s what you get when you build a system on top of the chatbot.
What’s actually on her site now
Go look at vestaconsultinggroup.com. I’ll wait.
The neighborhood guides cover intown Atlanta in real depth. Buckhead, Decatur, Ansley Park, the rest. There’s a blog. There are market reports. There are videos. There’s a clean FAQ section. There’s something she calls the Vesta Method, which is her three-step selling framework laid out so a seller can read it and understand exactly what working with her looks like before they ever pick up the phone.
She has 4,000+ newsletter subscribers. Property search is built right in, no iHomeFinder needed, no separate platform. The brand identity is intentional. Female-owned, LGBTQ+ friendly, education-first. None of that is hidden. It’s the first thing you read.
And about three months in, she’s getting closings and listings from it. Her words on that, “still early.” That’s the right way to talk about it. SEO compounds. What you publish today doesnt rank tomorrow, it ranks six months from tomorrow. The fact that she’s getting deals already, with a system that’s still warming up, is the leading indicator.
The math nobody wants to admit
Lets do the simple math because the math is the part that should bother every agent reading this.
iHomeFinder Optima Express, $150 a month. The SEO platform she was about to sign up for, $350 a month. That’s $500 a month she would have been paying, on top of the developer she’d already paid, forever. $6,000 a year. $30,000 over five years. And she didnt sign up for the second one and she canceled the first.
Now what does the AI stack cost? A Claude subscription is $20 a month. If you want the higher tier, $200 a month. Even at the high end, it’s a fraction of what the legacy stack runs.
The tools that used to require $500 a month plus a developer on retainer now require an afternoon and a subscription that costs less than the gym you dont go to. That is the part of this I dont think most agents have processed yet. The economics of running a real estate business changed and the bill hasnt arrived for most of the industry.
Why this article isnt really about Ariella
I’ve been thinking about this for months because I’m seeing a pattern. There are two camps of agents right now.
Camp one is paying for the developer, paying for the IDX feed, paying for the SEO platform, paying for the lead service, paying for the CRM with the AI add-on, and waiting for somebody else to make their phone ring. That’s the camp the industry built around. The vendors love that camp.
Camp two figured out that they can build the same stack themselves, in their voice, on their domain, and own all of it. That’s a smaller camp. It’s growing. Ariella is in it.
The frustrating thing is that most of camp one knows camp two exists. They’ve heard about it. They’ve maybe tried ChatGPT once and it gave them a generic blog post and they wrote it off. They didnt build the voice profile. They didnt build the system. They tried the chatbot and stopped.
You dont have to be technical to do what Ariella did. She isnt a developer. She’s an agent who decided she was tired of paying for things she could build. The technical part is genuinely a few weekends of work and a willingness to figure things out. The hard part is the decision.
What I told her, what I’d tell you
When Ariella reached out, I told her two things. First, build the voice profile before you build anything else, because if your AI doesnt sound like you it doesnt matter how good the content is. Second, dont try to replicate every feature of your old site on day one. Ship a smaller version that you actually own and add to it. The mistake people make is trying to recreate the legacy stack one for one and they get overwhelmed and quit.
She nailed both. She also did something most people dont, she committed before she knew it would work. She canceled the iHomeFinder subscription. She didnt sign the SEO contract. She put real money on her own ability to build, which is the thing most agents wont do because it feels safer to keep paying.
(That instinct, by the way, the “let me keep paying so I have someone to blame if it doesnt work” instinct, is exactly the trap Kahneman writes about in Thinking Fast and Slow. We hate loss more than we love a gain of equal size, which is why we keep writing checks for stuff we could replace ourselves. It’s a bias, not a strategy.)
If you want to read more about how agents are actually using AI in their day to day work, including some specific Claude workflows that go beyond just blogging, take a look at the AI for Realtors guide on this site. It’s a good companion to this story.
The 90 day version of Ariella’s story is that she saved $6,000 a year, took control of her platform, and is already getting business from it. The 5 year version is that she will be running a media business that happens to sell real estate, while a lot of her competitors will still be paying their developer to swap a hero image. I know which one I’d rather be.
Frequently Asked Questions
The takeaway
Most agents are one decision away from owning their platform instead of renting it. Ariella made that decision 90 days ago and saved $6,000 a year while getting better results. The tools are out there. The math is out there. The only thing missing is the agent willing to put a weekend on the calendar and start.
If you’re working on something similar in your own business and want to talk it through, you know where to find me.
