ATTOM has added an AI-powered model to its rental valuation product and now says it can estimate the market rent for nearly 100 million U.S. homes, including many that have never been rented. The property data company announced the upgraded Rental AVM on Nov. 3, 2026. It says coverage grew by about 40%, and that the model’s median error is 5.9% on homes before they are listed for rent.
An automated valuation model, or AVM, is software that estimates a property’s value from data rather than an appraisal. Sale-price AVMs are common. Rent is harder to estimate, and ATTOM’s own CEO named the reason in the company’s announcement.
“Estimating rental values is difficult because leases are private and many homes have no rental history,” said Rob Barber, CEO at ATTOM. Unlike a sale price, a lease is usually not part of the public record.
How ATTOM’s AI rental model works
According to the Irvine, Calif.-based company, the new model runs several machine-learning models that look at a home’s characteristics, its estimated value, any rental history and recent rental activity nearby. It weighs each input according to how much evidence exists for that property. It is trained and benchmarked on observed rental listings, which ATTOM says it cleans to remove duplicates, placeholder rents and unrealistic prices.
Each estimate comes with a low and a high value. ATTOM says that range is calibrated and narrows as the evidence behind an estimate grows. In principle, a home with its own rental history and plenty of nearby listings should get a tighter band than a house in an area with little rental activity.
“Our enhanced Rental AVM evaluates multiple sources of rental information and adapts to the evidence available for each property,” said Aaron Wagner, head of data science at ATTOM. He said that adaptability is what lets the model estimate rent for homes with no rental history and still keep the ranges tied to the strength of the evidence.
Estimates cover single-family homes, condominiums and townhouses in all 50 states and the District of Columbia, and the estimates update monthly.
Who uses rent estimates
ATTOM is marketing the product to three groups. Investors and single-family rental operators can use it to screen acquisitions and set rents. Mortgage lenders and insurers can use it to assess rental income on investment-property loans, including debt-service-coverage-ratio (DSCR) loans that qualify borrowers on a property’s rent rather than their personal income. Portals and AI developers can build the estimates into listings and investment tools.
The model is available through ATTOM’s APIs, bulk data licensing, the Snowflake and Databricks data platforms, and its MCP Server. MCP, or Model Context Protocol, is a standard that lets AI assistants and agents pull data from outside systems. That makes the rent estimates available to AI tools as well as to analysts.
The release continues ATTOM’s push to make its data usable by AI software. In August the company launched three AI agents that answer plain-English questions about its property data, under the ATTOM Intelligence banner Barber referred to in the announcement. ATTOM says its broader database covers more than 160 million U.S. properties. It also publishes market research that RealtyWire covers, including a recent report on home-flipping margins.
ATTOM did not publish accuracy figures for homes that are already listed or have rental histories, or comparisons with competing rent models. Buyers will need to test the 5.9% median error against their own markets. More coverage of AI tools in real estate is on RealtyWire’s technology and AI page.
