🎧 We’re also on Apple Podcasts and Spotify!

Three weeks after Meta launched its personal assistant, Muse reportedly reached 3.4 million downloads and the number-one spot in the app store. It can search, compare, schedule, purchase, and continue working after the app is closed.

That means your next buyer may not begin with Zillow, Realtor.com, or Google. They may simply say:

“Find me a three-bedroom home under $600,000 and schedule the showings for Saturday.”

The question is no longer whether AI can help consumers search for homes. The question is whether AI can find you.

This week's PowerHouse Talk covers Meta Muse, AI-generated inspections, pricing analysis, virtual staging, expired listings, direct mail, and the creative work agents will have to do once busywork disappears.

Tammy opened the show with a simple, old-school prospecting tactic: the golden letter.

The letter contains one sentence:

Tammy mailed the letters with a real signature and a real phone number. One seller called, she signed the listing, arranged professional photography, and prepared an open house.

Then she handed the listing to AI. The system created the Facebook post, scheduled stories, and prepared the open-house marketing while her assistant Andrea was freed up to do more valuable work.

The lesson is not that old-school marketing beats AI or that AI beats old-school marketing. The combination is the point:

Dan described a listing opportunity involving a roughly $3.5 million home. The seller had a handshake arrangement with another agent: if that agent found an off-market buyer willing to pay $3.5 million, the seller would consider selling.

Dan could have accepted the same arrangement. Instead, he challenged the logic. Without exposing the property to the market, how could anyone know whether a buyer might pay $3.6 million or $3.7 million?

His recommendation was to properly market the property rather than quietly accept the first possible buyer. That protects the seller's interests and creates a genuine price-discovery process.

The seller's motivation was also important. The house was too large now that the children were grown, and the seller wanted to downsize and begin estate planning. This was not simply a pricing conversation. It was a life-transition conversation.

That is where AI can help research and prepare — but the agent still has to understand why the seller is moving.

The show introduced a growing library of interactive resources being built for Libertas and EXP agents. These are not static PDFs. They are interactive mini-websites designed to walk agents through specific skills and systems.

The resources discussed include:

The larger strategy is to combine training, AI, and coaching so agents do not have to figure out every workflow alone.

Julie described a growing transaction problem: buyers are feeding inspection reports into AI and asking what they should request from the seller.

The AI often produces an extensive repair list containing every possible issue. The buyer then sends the entire list to the buyer's agent, who sends it to the listing agent. What used to be a difficult inspection conversation becomes an enormous demand package that can threaten the deal.

Julie said the problem is partly that AI does not understand priorities. It may not know the difference between:

The recommended solution is to get ahead of the problem through a pre-inspection. A seller can identify major issues, collect invoices, address safety and security concerns, and present the home in the best possible condition before the buyer's inspection happens.

The buyer's agent still needs to use AI carefully. A better prompt should ask AI to rank inspection findings by:

The AI can help organize the report, but it should not replace professional judgment.

Tammy's advice was simple: even if the agent is not using AI for every part of a comparative market analysis, the seller probably is.

That means agents should understand what AI is likely to tell the client.

One practical use is to give AI the comparable-sales list and ask it to identify differences in:

The agent still chooses the pricing recommendation. AI checks the work, identifies gaps, and helps explain the analysis.

Tammy also recommended giving sellers a written value and net-proceeds sheet. The purpose is not to hide behind AI. It is to make the agent more prepared and more transparent.

Meta Muse reportedly went live in September and quickly reached 3.4 million downloads. The show describes it as a consumer-facing assistant that can:

Tammy tested it by asking for three homes in Herndon, Virginia, priced between $600,000 and $700,000 with at least three bedrooms and two bathrooms.

Muse returned three properties and explained the advantages and disadvantages of each. It also provided:

Tammy noticed that Muse used Redfin when Zillow did not provide the information it needed. That suggests that the AI assistant may use multiple portals as data sources while presenting the consumer with a single conversational interface.

The consumer may never visit the portal directly.

The panel's advice is to make sure your public information is complete, accurate, and easy for AI systems to understand.

Agents should publish:

A profile hidden inside a portal or brokerage system may not be enough. AI needs public, searchable evidence that explains who you are and what you do.

The listing agent also needs to understand how AI describes the property. If the system tells buyers that a home's assigned schools score two out of ten, the agent should know that before a consumer brings it up.

Chris described using another AI assistant as a highly capable personal productivity tool. It can find old emails, locate information, and connect data across systems.

The benefit is obvious. The risk is also obvious: the assistant may have access to email, passwords, financial information, property records, and other sensitive accounts.

The panel compared that concern with trusting human employees and property managers who already have access to many of the same systems. The difference is that AI platforms may be targeted by people whose primary mission is to gain access to large amounts of data.

Before connecting an AI assistant to business accounts, agents should consider:

The productivity gains are real, but access should be designed rather than granted automatically.

Dan discussed a Reddit thread with more than 1,100 upvotes criticizing AI-generated staging. Buyers complained that some images changed the structure of the home, created unrealistic furniture layouts, or showed rooms that could not physically exist.

One example placed a dining table at the top of a staircase. That is not a minor editing issue. It signals that no one checked the output.

The panel's position was nuanced:

Chris argued that virtual staging can be useful because many buyers struggle to imagine the spatial relationship of an empty room. Julie added that the first rule is simply not to be lazy. Agents need to inspect every image and ensure that it makes sense.

The dividing line is deception. Showing what a room could look like is different from pretending that the room already looks that way.

Chris shared one of the most important personal insights of the episode. At a previous company, he worked from early morning until night with only short breaks. After moving into a role that limited his schedule, he discovered something unexpected: he finally had time to think.

He began having better ideas, more productive thoughts, and more creative solutions because his entire day was no longer consumed by execution.

That is the real opportunity with AI. It does not merely help agents do more tasks. It can create enough space for agents to think about:

But the extra time has to be protected. If agents simply fill it with more notifications, more meetings, and more scrolling, the benefit disappears.

The panel challenged each other to identify what AI cannot do.

The answers included:

Chris said real estate is only the game he chose to play, not his identity. That distinction matters because AI may remove many tasks previously used to define professional value.

If your identity depends on being the person who keeps the spreadsheet perfect, answers every administrative question, or remembers every detail, AI may feel threatening. If your identity is broader — mentor, parent, creator, investor, coach, community member — AI can give you more room to live those roles.

Chris's answer was direct: if the home has been on the market for 60 days, the price is too high.

The broader conversation should begin with motivation. Ask whether the seller still wants to move, where they want to go, and what deadline matters. Then explain that the market has provided feedback.

A useful framing is:

The agent's job is not merely to request a price reduction. It is to connect market positioning to the seller's actual goal.

Dan's answer: consider canceling the Zillow spend immediately.

If the close rate is approximately 3%, an agent may need around 33 leads to generate one closing. At $3,000 per month, that could mean spending roughly $12,000 to $15,000 to create one transaction, depending on referral fees and closing rate.

Dan recommended redirecting the money toward:

The episode cites a market where more than 5,000 expired listings above $500,000 were available since the beginning of the year. The issue may not be a shortage of leads. It may be that agents have been trained to buy leads instead of creating relationships.

Tammy recommended using AI as a checking and explanation tool rather than handing it complete authority.

Give AI the comparable-sales list and ask it to flag differences, identify inconsistencies, and explain why one comparable may be more or less relevant. Then make the pricing decision yourself and present the analysis with a clear net-proceeds sheet.

That approach lets AI strengthen preparation without replacing the agent's responsibility.

The next wave of real estate will not be defined by who has access to AI. Everyone will.

It will be defined by who uses AI responsibly and creatively:

Meta Muse may find the properties. AI may compare the homes. A bot may schedule the appointments. A model may analyze the inspection report.

But someone still needs to help the client decide what to do.

That person will be the agent who can combine technology with judgment, trust, creativity, and human connection.

— Tim, Julie, Dan, Keri Chris, Kacie, Tammy and Orlando
Hosts, Power House Talk

whylibertas.com/harris
whylibertas.com/dan
whylibertas.com/heller
whylibertas.com/kacie-anderson
whylibertas.com/orlando-montiel

💬 Want your question featured on an upcoming episode? Reply to this email or DM us on Instagram— we might break it down next.

📩 Know an agent who should be in this room?
Forward them Power House Talk.