US real estate agents have adopted AI. Here's what it does for their productivity.

Almost half of US Realtors now use AI every week. In the National Association of Realtors’ 2026 Technology Report, released September 22, 23% use it daily and 25% weekly, up from 20% and 22% in 2025. Another 31% experiment with it occasionally. The rest don’t use it: 9% are curious but haven’t tried it, and 12% have no plans to. That is 21%, against the 32% who had never used AI in their business in 2025.

A February 2026 survey of 225 agents by Realtors Property Resource (RPR), NAR’s data company, found that 82% use AI and 68% use it daily or several times a week. The two surveys use different samples, but both find about four in five US agents using AI.

We haven’t found a Canadian survey of agents at this scale, so the US numbers are the closest data we have for Ontario.

What are US agents using AI for?

Mostly writing. Among AI users in NAR’s 2026 report:

Task	Share of AI users
Listing descriptions	75%
Social media posts	56%
Emails and follow-ups	52%
Market summaries	30%
Reviewing or summarizing documents	27%

ChatGPT is the main tool. Chicago Agent Magazine’s coverage of the report says ChatGPT use rose from 58% to 91% in a year.

Are agents who use AI more productive?

Measured in time saved, the surveys say yes.

RPR, published February 2026: 68% of agents said AI saves them at least one hour a week, and 34% said four hours or more. 71% name time saved as AI’s main value.
Federal Reserve Bank of St. Louis: in its tracker, updated in August 2026, the share of all US work hours saved by generative AI, counting workers who don’t use it, grew from 1.6% to 2.2%.
National Bureau of Economic Research: in a controlled study of 5,179 customer support agents at one company, those given an AI assistant resolved 14% more issues per hour. The least experienced improved by 34%, and the most experienced barely changed.

For the third of agents saving four hours or more, that is half a working day each week, or about 200 hours over a 50-week year. And if you’re newer to the business, the support-agent study found the largest gain among the people who had done the work least often.

Does the time saved turn into more deals?

Nobody has measured that yet. No US survey we found compares closed transactions for agents who use AI against agents who don’t.

The closest measure is how agents rate the effect on their business. In NAR’s 2026 report, 55% said AI has had a positive effect, with 39% calling it moderately positive. 34% said it made no noticeable difference, down from 46% in 2025. For a third of agents, AI hasn’t changed their business yet.

The task list suggests why. Listing descriptions, social posts and emails are real work, but they are marketing. They don’t move a signed offer closer to closing. The work that does (drafting the Agreement of Purchase and Sale, filling in the schedules, collecting signatures, keeping FINTRAC records complete) is lower on the list or absent. Document review is at 27%.

RPR’s survey shows the hesitation. Agents’ top concerns about AI were the accuracy of its output (63%), compliance and legal issues (49%), and misreading market data (47%). A general chatbot doesn’t know which form a clause belongs on, and it states a wrong date as confidently as a right one. Agents are right not to hand it a legal document.

Which work should AI take off your desk?

The offer is the most repetitive part of the job: the same names, address, legal description, price, deposit and dates entered into the Agreement of Purchase and Sale, its schedules and the forms around it. It is also where a typo costs the most.

That is the work OfferCopilot takes on. It drafts the Form 100 Agreement of Purchase and Sale and its schedules from the listing and your client’s file, and fills the forms from that structured data, not from free-form AI output. AI suggests clause wording and reads data out of listings and uploaded documents, and you see every suggestion before it’s saved. Signing is built in, and nothing goes out without you. We worked through what that saves in what your offer paperwork really costs.

What should Ontario agents take from the US numbers?

Whether to use AI is mostly settled. 88% of US Realtors either use AI or are curious about it. Holding out is a choice most of your competitors aren’t making.
Track what you do with the hours. If AI saves you four hours a week on emails and you spend them writing more emails, your business hasn’t changed.
Use AI on the work tied to closing, with a tool built for it. Keep your own judgement on every number that goes on a form.

For more on where AI fits in an agent’s practice and where it doesn’t, read our earlier post on AI for real estate agents, or see the FAQ for what OfferCopilot does and doesn’t do.

Start your free trial → Use code FREE-TRIAL for your first 2 months free.

Frequently asked questions

What percentage of US real estate agents use AI?

In NAR’s 2026 Technology Report, 23% of Realtors use AI daily and 25% weekly, up from 20% and 22% in 2025. Another 31% experiment with it occasionally, and 21% don’t use it.

How much time does AI save a real estate agent?

In RPR’s survey of 225 US agents, published in February 2026, 68% said AI saves them at least an hour a week and 34% said it saves four hours or more.

Do agents who use AI close more deals?

No US survey has measured that yet. In NAR’s 2026 report, 55% of agents said AI has had a positive effect on their business and 34% said it made no noticeable difference. Most agents use AI for marketing, such as listing descriptions and social posts, rather than for the offer itself.

What do real estate agents use AI for?

Mostly writing. Among AI users in NAR’s 2026 report, 75% use it for listing descriptions, 56% for social media posts and 52% for emails and follow-ups. 27% use it to review or summarize documents.