A typical $400,000 home sale in the United States carries about $39,660 in transaction costs, and a new report from Alloy Advisors says half of that is overpriced. The same report, covered by Real Estate News on June 11, found that agents clearly beat AI on only 3 of the 23 tasks a sale actually requires. That is the week in one sentence. The tools your competitors are testing are no longer aimed at saving you an hour. They are aimed at the line items that pay your mortgage.
Here is the week for the operator who was too busy to read it as it happened. Five things moved, one of them loudly. We pulled the signal and added our read on what to do before Monday.
The number that should reset every commission conversation
Start with the Alloy report, because it reframes everything else. The argument is not that AI replaces agents. It is that AI collapses the price of the work agents have always charged for. When software can run comps, draft disclosures, answer buyer questions at midnight, and schedule showings, the 23-task bundle that justified a 5 percent commission starts to look like three tasks that justify a much smaller one. The report names those three durable tasks: in-person negotiation, local judgment that is not in any dataset, and the trust a client extends to a face they know.
Our read: stop defending the commission and start itemizing the three tasks you win on. The agents who survive this will say, out loud and on their listing page, exactly what they do that a model cannot. The ones who lose will keep quoting a percentage and hoping nobody does the math. A buyer with a free AI co-pilot will do the math. We covered the buyer-facing version of this shift on Tuesday, when Realtor.com’s RealAssist started answering buyers before their agent could. The counter-move we laid out is the same instinct applied earlier in the funnel: be the answer, or watch the machine become it.
Agentic software stopped being a demo
RISMedia ran a feature on June 11 arguing that real estate tech has crossed from reactive tools to agentic systems that own whole workflows. The clearest example is Lofty AOS, the operating system Lofty shipped in February that now runs named agents for lead qualification, social posting, and seller-lead hunting. The point for an operator is not the product. It is the shape of the thing. Vendors are no longer selling you a feature you click. They are selling you a worker you supervise.
That changes the buying question. You are not asking “does this save me time on task X.” You are asking “what is this agent allowed to do without me, and how do I see what it did.” Any tool that cannot answer the second half of that question is a liability, not a hire. Approve the rules, keep the audit trail, and treat the agent like a new employee on probation, not a magic button.
AI Overviews now own a quarter of search, and Google said the quiet part out loud
Two search items landed together this week. First, AI Overviews now appear on more than 25 percent of US searches, and several trackers report click-through dropping by more than half on broad informational queries. Second, on June 5 Google updated its own Search Central documentation to state plainly that GEO and AEO are not new disciplines. They are SEO. The same docs added a section telling local and ecommerce businesses to feed Merchant Center and keep their Google Business Profile current if they want to show up inside AI answers.
We argued back in May that Google would fold GEO back into plain SEO, and now it is in writing. The takeaway has not changed, it has hardened. There is no separate AI playbook to buy. There is your profile, your structured content, and whether a model can quote you cleanly. On Wednesday we showed how your Google Business Profile now reports an AI-surface share, and why anything under 15 percent is a problem you can see. If you read one thing from this brief, make it that one, then go check your number.
The small-business adoption data finally got specific
Two data drops this week cut through the hype. The SBE Council reported on June 5 that 82 percent of small employers have adopted at least one AI tool, with a median of five tools per business. A Small Business Expo survey published June 9 found that 78.9 percent of owners say AI is more useful than a year ago, while 63.7 percent feel little or no pressure to adopt it. Read those two together and you get the real story. Adoption is high and rising, but it is pull, not push. Owners are not scared into it. They are buying it because it pays.
The same survey found the top marketing use of AI was research at 28.8 percent, ahead of content creation. That matters because most operators still think of AI as the thing that writes social posts. The operators getting paid are using it to decide, not just to draft. The trades version of this showed up on Thursday, when we ran the numbers on dispatch. Most service trucks bill only about 70 percent of the day, and AI dispatch closes that gap by optimizing for revenue per route instead of raw efficiency. That is a decision tool, not a chatbot, and it is where the money is.
What to do before Monday
Five items, five minutes each. Pick the ones that fit your trade.
- If you sell homes: write the three tasks you do that a model cannot, and put them on your listing presentation this weekend. Stop selling the percentage. Sell the three things.
- If you run local marketing: open your Google Business Profile insights and find your AI-surface share. If it is under 15 percent, your profile is stale or your content is unquotable. Fix one this weekend.
- If you run a service trade: pull last week’s dispatch log and calculate billable hours as a percentage of paid hours. If it is near 70 percent, you have a five-figure leak that routing software can plug.
- If you are evaluating any agentic tool: ask the vendor exactly what the agent can do without your approval and where you see its actions. No clear answer, no purchase.
- Everyone: name one decision you made last week on gut that you could have made on data. That is your first real AI use case, not your next blog post.
At Atlas Unchained we build the websites, automations, and small tools that let an operator act on a brief like this without hiring a department to do it. If any item above turned into a “we should really fix that,” that is the work. Tell us what is leaking and we will help you plug it.
One last thing before you close the laptop. The commission report is really a question pointed at every trade, not just real estate. So here it is. If a client could buy the routine 80 percent of your service from a machine for a fraction of your price, would they still pay you for the 20 percent only you can do? If the answer is yes, write down what that 20 percent is, because it is the whole business now. If the answer is no, you have until your competitor figures it out to change the answer. Have a good weekend. Read you Monday.
About the Author
Trevor Kaak is the founder of Atlas Unchained, a portfolio of products and services helping local businesses run leaner with AI — from custom websites to vendor-bidding marketplaces to vertical SaaS. He writes about marketing, automation, and the craft of building software for operators who’d rather work on their business than in it.