AI Case Studies  /  Case study

Does AI Push Real Estate Commissions Lower?

A June 2026 Alloy Advisors report argues that AI erases much of the work bundled into a listing fee. Here is what the economics actually say for agents, and where the human premium still holds.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 2, 2026

AI is beginning to press on real estate commissions, mostly by giving consumers tools that used to justify part of the fee. In "The Home Sale Transaction, Reconsidered," Alloy Advisors (Amit Kulkarni and Russ Cofano) estimates that the routine tasks in a listing, such as comparative market analyses, MLS entry, listing copy and offer modeling, now cost close to zero for a competent AI user, for little more than the price of a monthly AI subscription. Portals and brokerages including Zillow, Redfin, Compass and The Real Brokerage have shipped AI tools that do exactly this work, and at least one seller has already used a chatbot to skip the listing agent entirely. Commissions have not fallen yet, but the value case behind them is changing.

The report that put a number on the fear

For years, agents heard that technology would compress commissions "someday." A June 11, 2026 report from Alloy Advisors, the consultancy founded by industry veterans Amit Kulkarni and Russ Cofano, tries to put dollars on that someday. In The Home Sale Transaction, Reconsidered, the authors take a benchmark $400,000 sale and count $39,660 in hard transaction costs, about 9.92% of the price. Commissions are the biggest single piece at roughly $23,000, or 5.75% of price and 76% of all friction the seller pays. The rest goes to transfer taxes, title insurance, settlement and loan fees, appraisals and inspections.

The report's sharpest claim is not the total. It is the split. Kulkarni and Cofano argue that AI now serves the consumer, not just the agent, and that AI-equipped buyers and sellers can see, question and negotiate individual line items in ways they could not a few years ago. That is the mechanism that could move fees.

What the software has already eaten

The report separates agent tasks into two tiers, and the first tier is the one under threat. Tier 1 covers comparative market analyses, MLS entry, listing descriptions, offer modeling, transaction coordination and basic disclosure checks. Pre-AI, Alloy pegs that bundle at roughly $1,500 to $3,500 per listing. With current tools, they estimate a seller with well-prompted AI can replicate the core of that work for the cost of a monthly subscription.

This is not a forecast. It is a description of tools that already shipped. Redfin launched a conversational home search built on Sierra's AI platform in November 2025, letting buyers refine a search through natural dialogue instead of filters, a job that used to route through an agent. The Real Brokerage introduced HeyLeo, a voice concierge for buyers, alongside its Leo CoPilot agent assistant. Zillow's CTO David Beitel told Fortune that AI is reinventing every step of the buying process, and Zillow has even turned Google's NotebookLM into a buyer marketing channel. When the portals hand consumers the Tier 1 work for free, an agent charging for it is selling something the client can get elsewhere.

The tools point inward too

The same AI wave is landing inside brokerages, which cuts both ways. It lowers the cost to run a deal, and it makes the "we do so much behind the scenes" argument easier to prove or harder to hide. Compass shipped Compass One, a client dashboard, in early 2025, then added a voice-activated AI assistant that drafts emails, builds follow-ups and creates marketing collateral. MoxiWorks launched RISE, a marketing platform that predicts client intent and automates follow-up. Rayse, an agent value platform, partnered with Columbus Realtors to give more than 9,000 members free access to a tool whose entire pitch is showing clients, in real time, the work an agent actually does.

Read that last one carefully. When the industry itself builds software to prove agent value, it is conceding that the value is no longer obvious. That is the honest subtext of this whole cycle, and it matters more to your pricing than any headline about AI listing bots.

Why commissions have not actually dropped

Here is the part agents should sit with, because it cuts against the doom narrative. Commissions have not fallen. Citing Clever Real Estate and Redfin data, Alloy notes the national average commission rose to 5.44% in mid-2025, up from 5.32% a year earlier. A HousingWire agent survey from April 2025 found 58.8% of agents saw no change in buy-side commissions after the NAR settlement took effect, and 11.76% said theirs went up. Rates have been sticky.

The report gives structural reasons: the practice changes did not remove seller-paid buyer commissions in reality, 13 states and Washington, D.C. effectively ban à la carte brokerage, and agents only get paid when a deal closes. So the settlement, the thing everyone braced for, is not the force compressing fees. If compression comes, the report argues, it comes from AI-assisted consumers and new pricing options first, long before regulators or pure AI listing platforms reshape anything.

The trust gap is your runway

The reason this is a pressure story and not a collapse story is trust. That December 2025 YouGov survey is the number to hold onto: 65% of Americans trust AI to compare prices on major purchases including homes, but only 14% trust AI to act on their behalf in those decisions. Realtor.com research cited in the report found about 82% of active buyers and sellers already use AI for housing insights, and they rate agents and AI nearly evenly on which makes them "smarter," with agents still seen as more accurate.

So consumers will increasingly arrive informed. They will know the comps, the line items and the range of reasonable offers. What most of them will not do is negotiate their own biggest transaction or carry the legal accountability. As Cofano put it, people dislike buying cars for the same reason, and the agent as active negotiator and trusted companion is still worth money. That gap between "informed" and "willing to act alone" is the runway. It is real, and it is not infinite.

Reposition before the market repositions you

The strategic move is not to defend the old bundle. It is to reprice around the part AI cannot commoditize. A few honest implications for anyone selling real estate skill for a living, and the ones we teach inside [The Leveraged Years real estate track](/leveraged-real-estate):

First, price the human core out loud. Alloy values negotiation, on-site judgment, hyperlocal knowledge and fiduciary accountability at $2,000 to $6,500 per transaction, and it does not scale with home price. If your fee is a flat percentage, be ready to say what the client is buying at the top of the range beyond a number that happens to be bigger.

Second, use the same AI your clients use, faster. The agents who win are not the ones who avoid these tools. They are the ones who run Tier 1 in minutes and spend the reclaimed hours on the negotiation and relationship work that still moves outcomes. That is the pattern across every [AI case study we track](/ai-case-studies/), and it echoes what we see in [how small businesses run on AI](/how-small-businesses-run-on-ai): the tool absorbs the routine, and the operator moves up the value chain.

Third, take the do-it-yourself path seriously, because a few clients will. Stuart Thompson, a New York Times reporter, used Google Gemini to write his listing, find a photographer, stage the home and post to the MLS, then cleared about $90,000 more than expected on the sale. That is still an edge case, and it took a confident, technical seller. But it is no longer hypothetical, and pretending it is will not help you.

The great agent still wins, as the report concludes. The difference in 2026 is that "great" now has to be visible and specific, priced to the work only a person can do, because the rest is already free.

Browse all AI Case Studies

Informational analysis for working professionals, not professional advice.