Real Estate Operations
How I automated my client updates and got my Sundays back
- real-estate-agents
- ai-tools
- client-communication
- listing-workflow
- productivity
Every Sunday night for six years, I opened the same three Google Docs.
Not because I wanted to. Because if I did not, Monday would eat me alive. Each doc was a client. Each client needed an update by Tuesday morning. And every Sunday I was the only person in the world who knew what had happened that week on their file.
Inspection came back. The buyer called crying. The lender wanted one more pay stub. The listing agent on the other side ghosted for three days and finally texted at 9pm on Saturday.
None of that lives in a CRM. It lives in Gmail. In text threads. In showings apps. In my head.
So Sunday night I sat down at my dining room table with a glass of something and reconstructed the week for each client. Then I opened a fourth doc, called it Client Updates, and rewrote what I had just remembered into something my clients could actually read.
That was my Sunday.
The manual state was a filing cabinet made of guilt
I had a Google Sheet with 26 columns per client. I had a notes app that I opened during showings and never opened again. I had a whiteboard in the closet by the front door. I had sticky notes on the dashboard of my truck.
If you cut me open, you would have found a spreadsheet.
None of it worked. The Google Sheet was three weeks behind because I only updated it on Sunday nights when I was already tired. The notes app was full of one-line reminders that were meaningless out of context. The whiteboard was the last inspection I did. The sticky notes were the last three showings.
The actual state of any given client lived in my head. And my head is a real estate agent's head, which means it also holds the school district for a house I sold in 2021, the name of a lender who ghosted me in 2019, and the door code for a listing that expired eight months ago.
That is not a system. That is a personality disorder.
The moment I saw Claude do it
The first time Brett showed me what Claude could do with my actual Gmail, I almost got up and left the room.
He pointed it at seven days of email for one active listing. Told it to draft a client update in my voice. Referenced my last three updates for tone. Named the specific facts it should include and the specific ones it should leave alone.
The draft that came back was the update I would have written on Sunday night. It was not better than mine. It was mine, without the two hours of me sitting at the dining room table looking at a screen.
The specific moment I remember is when it flagged the ghosted-listing-agent thing without me asking. It said something like, hey, the listing agent on the other side has not responded in three days. Do you want to mention that or keep it internal. That is a decision I make every Sunday. Claude just surfaced it and let me make the call.
That was the point where the question flipped. It was not, can this work. It was, why am I still writing these by hand.
We built it and named it Tuesday Updates
BNB AI Solutions was born out of that Sunday. Brett and I sat down and built the first Claude-powered task together. We called it Tuesday Updates because that was the deadline it existed to hit.
Here is what it actually does.
Every Tuesday morning at 8:05am, it wakes up. It reads the last seven days of Gmail for each active listing. It knows which threads belong to which client because I labeled the accounts once, at the start. It pulls the facts. It writes a draft in my voice, using my last three updates as the tone reference. It flags anything it is unsure about with a note in the margin.
Minutes later, drafts are sitting in a Google Doc waiting for me. I read them, correct the one thing that was slightly off on any given draft, and hit send.
Total time on client updates for the week: fifteen minutes.
Total time before Claude: three hours on Sunday plus the anxiety tax of knowing it was coming all weekend.
The specific thing that made it work: my voice
There is a version of this task that would not have worked, and it is the version that writes generic real estate agent copy.
Every agent has voice tics. Mine include starting with "Quick update" when nothing is on fire, and starting with "Heads up" when something is. I always end with a small ask or a next action, never a signoff. I use "we" when I am talking about the transaction team and "I" when I am talking about what I did that week. I never say "circling back" and I never say "at the end of the day."
If Claude wrote client updates that read like Zillow marketing copy, my clients would have caught it in one email. What we did was point the task at my last two months of client updates and tell it to match the pattern.
The draft it produces now is closer to my writing than a lot of my own drafts on tired days. My wife read one last month and asked when I had time to write it. I told her Claude wrote it. She said it sounded like me. I said that is the whole point.
That voice-capture step is the part that makes this a real install and not a demo. It is also the part I would not have figured out by myself. Brett built that piece and it took two calls to get right.
The pushback I expected and never got
I expected two flavors of pushback and got neither.
The first was from clients. I was ready for someone to say, hey, did you write this or did a robot write this. Not one client has asked. The reason is that the update sounds like me because it is trained on me. It uses my phrases. It ends with the same kind of small sign-off I always use. The facts are correct because the facts came from my actual email.
The second was from other agents. I told my team lead I was doing this. He asked one question. How much time. I said fifteen minutes a week. He asked me to install it for him.
The industry likes to talk about clients not wanting AI touching their transactions. In my experience, clients want an agent who is present, prepared, and not falling behind. They do not care where the prep came from. They care that the answers are right and they arrive on time.
What Sunday looks like now
Sunday afternoon, I take my kid to the trampoline park.
Sunday night, my wife and I watch a show. I do not open a laptop. If something urgent happens between now and Tuesday morning, the Morning Brief task will surface it. If a new lead comes in from the site, Prospect Call Prep will draft the research before the call.
The three hours I was spending on Sunday client updates went back into my life. Which is what should have happened three years ago.
What this ships as
Tuesday Client Updates is one of the fifteen scheduled tasks that come with the Listing Machine install. Others include Morning Brief, Evening Recap, Weekly Strategy Review, Monthly Market Stats, and ten more. Alongside the tasks, the install ships three Claude skills triggered by natural language: Prospect Call Prep, Active Listing, and New Client. Everything runs on your own Gmail, your own Drive, your own Claude account, and your own Vercel deployment. Nothing sits on a third-party server we own.
If you want to see how the whole install works before buying, we have a full walkthrough on the How It Works page. The fuller version of the install story is in the Buddy Buck case study.
Two ways in from here. Listing Machine at $997, self-install, packages everything you need to install this yourself this weekend. If you would rather have us do the install for you, Listing Machine with Install at $4,997, typical turnaround under two weeks. If you want to talk to Brett or me first, use the contact form. We will walk you through your specific week and tell you honestly which path fits.
Written by
Buddy Buck
Co-founder, BNB AI Solutions · eXp Realty agent, Duvall WA
Practicing eXp Realty agent in Duvall, WA. Creator of Duvall TV. Host of the Buddy Buck podcast. Co-founder of BNB AI Solutions.
Founding cohort, first 10 agents
Install the Listing Machine PWA.
Full Progressive Web App, five Claude skills, deployed to your Vercel account. Yours to keep.
Related reading
AI-Native Operations
What an AI-native real estate business actually looks like
AI-native is not AI-enabled or AI-augmented. It is a business rebuilt from the ground up around AI as the operating layer. Here is what that looks like in a real real estate practice.
Brett K Moore · July 20, 2026
Real Estate Technology
Real estate agents do not need bigger CRMs. They need AI in their pocket.
The industry keeps shipping bigger CRMs to agents who already ignore the ones they have. What actually moves the needle is a small AI stack, on the phone, that answers when you talk to it.
Buddy Buck · July 20, 2026