Not long ago, I was reviewing the value of a South Florida condominium. One recently closed unit seemed to provide an obvious answer. It had sold for $265,000.

But the sale had taken nearly 300 days. The market had softened since it went under contract, and another unit was competing for buyers at a higher price with features the subject property did not have.

A computer could find the $265,000 sale in seconds. The harder question was how much that sale still meant.

That is the difference between finding a comparable sale and pricing a home.

Artificial intelligence is becoming remarkably good at gathering property information, organizing recent sales and detecting patterns. It can help an experienced agent work faster and examine the market from more angles.

What it cannot do is walk through a home, understand every difference between two properties, speak with the agents involved or accept responsibility when the market disagrees with its conclusion.

In Article Four, I explained why complete and accurate listing details matter. Those same details become critical when it is time to establish a price.

The Number on the Screen Is Not the Market

Homeowners now have access to more estimates than ever. With a few clicks, they can find an automated value, tax assessment, price-per-square-foot calculation and list of nearby sales.

Those numbers are useful starting points. They are not necessarily the price a buyer will pay.

An automated valuation generally works by identifying statistical relationships among available data. It may consider location, square footage, bedroom and bathroom counts, prior sales and nearby transactions.

The difficulty is that real estate data is rarely complete and homes are rarely identical.

The public record may show that two houses are approximately the same size. It may not show that one has a functional split-bedroom floor plan while the other has an awkward addition. It may record a swimming pool without explaining that one backyard feels private and the other backs up to a busy road.

A condominium model may appear identical on paper, but one unit may face the water while another faces a parking structure. One building may have completed its structural work and funded its reserves. Another may be facing a large assessment.

The number on the screen is an estimate produced from the information the system can see. The market includes everything buyers notice—even when the database does not.

What AI Does Well

Artificial intelligence can make the pricing process more disciplined when it is used correctly.

It can organize recent sales, active listings, pending transactions, price reductions and days on market. It can help compare properties by size, age, condition, location and documented improvements. It can also identify patterns that deserve a closer look.

For example, are renovated homes selling while dated homes remain active? Are buyers paying a premium for impact windows, newer roofs or stronger insurance profiles? Are listings between $700,000 and $750,000 receiving more attention than those just above $800,000?

AI can help an agent ask those questions earlier and test whether a pricing recommendation is consistent with the available evidence.

I also find it useful as a second set of eyes. After more than 22 years in real estate, I have seen many market cycles and property types. Experience is valuable, but it can also make any of us too comfortable with our first conclusion.

AI gives me another way to ask: What am I missing?

The answer still has to be checked. But the question can improve the work.

Closed Sales Tell the Past. Active Listings Reveal the Choice.

Most pricing conversations begin with closed comparable sales, and they should. Closed sales show what buyers were willing to pay and what lenders were willing to support.

But a seller is not competing only with properties that sold three or six months ago. The seller is competing with the homes buyers can purchase today.

That makes active listings important. They show the alternatives sitting on the buyer’s screen at this moment.

If a buyer can choose between two similar homes, the question is not simply what the neighbor received last year. The question is which home offers the stronger value now.

Pending sales provide another clue. They show where a buyer and seller have reached an agreement, although the contract price usually remains unknown until closing. Price reductions, showing activity and time on market also reveal how buyers are responding.

A strong pricing analysis therefore looks backward and forward. Closed sales provide evidence. Active competition provides context. Market direction tells us how carefully to connect the two.

The Most Similar Home May Not Be the Best Comparable

Selecting comparable properties requires judgment.

The house next door may not be the best comparison if it sold under unusual circumstances, needed extensive work or had features that materially changed its value. A slightly more distant sale may tell us more if it attracted the same type of buyer and offered a similar ownership experience.

The same is true with condominiums. Two units in the same building may differ because of floor height, exposure, view, condition, balcony, parking, assessments or the financial position of the association.

Price per square foot can help organize those sales, but it can also create false precision. Buyers do not purchase square footage by the pound. They purchase a particular home, in a particular condition, at a particular location, with a particular set of future expenses.

AI can line up the measurable differences. An experienced agent has to decide which differences buyers are likely to pay for.

From My Experience: One Sale, Several Different Meanings

The condominium I mentioned at the beginning illustrates the problem.

A nearby unit had closed for $265,000, but it took almost 300 days to sell. By the time I was evaluating another property, the market had moved. Buyers had more choices, and an active unit offered stronger features at a higher asking price.

An automated system could easily place heavy weight on the recent $265,000 closing. A seller could look at the same number and conclude that it established a floor beneath the property’s value.

I saw a more complicated picture.

The long marketing period suggested resistance. The changing market reduced the usefulness of an older agreement. The superior active competition placed a limit on how aggressive we could be. Condition, location within the building and the seller’s timing also mattered.

The sale was still relevant. It simply was not the answer by itself.

That is often what pricing requires: not rejecting the data, but understanding how much weight each piece deserves.

Pricing Is Also a Marketing Decision

Price does more than determine what a seller hopes to receive. It determines which buyers see the home, which competing properties appear beside it and what expectations buyers bring to the showing.

A property priced beyond the market may still appear online, but it may appear in the wrong comparison set. Buyers expecting more at that price will judge it against stronger homes. Buyers who might love it may never see it because their search ends below the asking price.

Overpricing can also weaken the most valuable part of the launch—the period when the listing is new and buyer attention is strongest.

A seller can reduce the price later. What cannot be recreated is the first impression among buyers who saw the home, rejected the value and moved on.

That does not mean every home should be priced below market. It means the price should be chosen deliberately, with an understanding of the competition, the seller’s goals and the probable buyer response.

The Market Gets the Final Vote

Even the best pricing analysis is a recommendation, not a guarantee.

The market may change. Interest rates may move. A competing property may enter the market. Insurance information may affect affordability. Buyers may respond differently than expected.

That is why pricing should not end when the listing becomes active.

Showing requests, online engagement, second showings, buyer questions and offers provide new information. If buyers are looking but not scheduling, the presentation may be the problem. If they are showing but not offering, the condition or price may be creating resistance. If there is little activity at all, the market may be telling us the property is positioned outside the buyer’s field of consideration.

AI can help organize those signals. The agent and seller still have to interpret them and decide whether the strategy should change.

The market gets the final vote. Good pricing means listening before that vote becomes expensive.

Gary’s Take

I do not believe artificial intelligence can price a home by itself.

It can find sales faster than I can. It can organize more information than any person could comfortably hold in mind. It can identify patterns, test assumptions and challenge an agent to explain why one comparable deserves more weight than another.

Those are meaningful advantages.

But a pricing recommendation still requires someone to walk through the property, understand the neighborhood, evaluate the current competition, recognize what the data does not show and have an honest conversation with the seller.

AI can organize the evidence.

Experience must interpret it.

And the market will ultimately decide whether we were right.

In Article Six, I will explain how AI can help identify the buyers most likely to value a particular home—and why successful marketing is not always about reaching the largest audience.