AI tools for real estate agents: 10 picks for 2026

What agents actually use AI for

These are 10 AI tools for real estate agents, matched to the jobs agents already use AI for. The National Association of REALTORS published its 2026 technology report on September 22. Among members, 23% use AI every day and 25% use it weekly. Another 31% are still experimenting. The share who are not using AI and do not plan to is 12%, down from the 32% who had not tried it in 2025. Those figures are in NAR's newsroom summary of the report.

Saving time is why 81% of members adopt new technology, up from 66% a year earlier. Improving the client experience is second, at 71%. Among agents who already use AI, the tasks are ordinary. Listing descriptions are 75%. Social posts are 56%. Emails and follow-up are 52%. Market summaries are 30%, and so is drafting marketing copy in a personal tone. Document review is 27%. Forty-one percent report using an AI content tool, and NAR's examples are ChatGPT, Copilot, and Gemini. The MLS is still at 96%.

Sixty-three percent said the learning curve is their biggest challenge with new technology, and 59% said cost. Thirty-six percent spend $50 to $250 a month on tech for the business.

Below are 10 tools, grouped here as writing, photos, and follow-up. NAR's AI uses are listing copy, social posts, and email. Photos are this post's grouping, not a line in that report. The usage numbers are NAR members, and several photo and data tools are built around US listing systems. Each product claim comes from the NAR release or from the company's own page. If the page and this post disagree, trust the page.

The 10 tools at a glance

Tools for listing copy and follow-up email

1. ChatGPT

NAR names ChatGPT as one of the AI content tools in that 41% group. Writing listing descriptions is the most common AI task among agents who use AI at all, at 75%. The workflow is plain. Paste the beds, baths, the street, and the three things you noticed on the walkthrough. Ask for a listing draft and a short email to the seller. The product is at chatgpt.com.

The failure mode is invention. A model will add a renovated kitchen because that phrase is common, not because you saw one. Read the draft against your notes. Cut any school, crime, or "who lives here" line you did not verify. You are responsible for the remark. The model does not know your board's rules.

2. Gemini

Gemini is Google's AI assistant, and NAR lists it beside ChatGPT. Google's Gemini plan page shows a free Gemini app. Gemini in Gmail is on the Plus, Pro, and Ultra plans. Docs is on Pro and Ultra, not on Plus. If the brokerage already works in Gmail, this is the assistant that can sit in the thread you are answering. ChatGPT only sees that thread if you paste it.

The editing rule does not change. NAR says 30% of AI-using agents write market summaries with these tools. A summary is only as good as the numbers you handed it. If you cannot show the client where a figure came from, delete the figure.

3. Microsoft Copilot

Copilot is the third name in NAR's examples, next to ChatGPT and Gemini. If the brokerage is on Microsoft 365, Microsoft Copilot is the assistant in that account. If it is not, start with ChatGPT or Gemini instead of adding another login. NAR's release names the product. It does not list Copilot features, so this post does not either.

Pick one of the three. They overlap, and the NAR figure is about using an AI content tool, not about paying for three drafts of the same remark.

Tools for photos, staging, and tours

4. Canva

Social posting is the second most common AI use in the NAR report, at 56% of agents who use AI. Canva's AI connector works inside ChatGPT, Gemini, Claude, and Copilot. You ask for a design, then edit the pieces. Canva says generated images can be split into layers in Gemini and ChatGPT. Live-data autofill is Canva Enterprise only.

Use it for the open-house card and the square crop of a photo you already took. It does not know the listing. You still type the address, the price, and the time.

5. REimagineHome

REimagineHome takes a photo of a room, including a furnished or cluttered one, and stages or redesigns it. The company says a transform takes about 60 seconds on average. You can follow up in plain language. Their example is "make the sofa black." They also say a detail can shift because the picture is generated, and that running it again is the fix.

Their FAQ answers the listing question in one line. You can put the images on sites such as Zillow or an MLS if you disclose them as "virtually staged" or "virtually enhanced." Put that label on the photo and in the remarks. A buyer who tours an empty room should not be surprised. I am leaving out the company's line about selling faster. That is their marketing, and we did not measure it.

6. Matterport

Matterport's Cortex AI builds a 3D digital twin from a capture. On that page they say it can calculate measurements, make 2D photos, recognize objects, and apply privacy protection. The same page describes Property Intelligence (square footage, ceiling height, wall dimensions, exportable to an MLS, in their words) and Project Genesis, which defurnishes a captured space or removes objects.

This is a different job from REimagineHome. One edits a still photo. The other starts with a scan. If you only have a phone album, you do not need Matterport yet.

7. Restb.ai

Restb.ai reads listing photos: room and feature tags, a condition read, photo-rule checks, captions, descriptions, watermarks, and duplicates. They say more than a million property photos are uploaded each day in the US. That figure is theirs.

It is not an app an agent downloads on a Tuesday. Restb.ai describes its customers as MLSs, portals, appraisers, and investors. If your MLS already subscribes, you may already get the tags. If it does not, this will not replace Canva on one listing. Messy photos still produce messy labels. More on that is in the data for AI guide.

Tools for follow-up, calls, and pricing

8. Lofty AI Sales Agent

Lofty's help center article, updated July 30, 2026, describes an assistant that works on three channels at once: chat on a Lofty website, text, and email. The job is to qualify a lead, fill in the profile, and hand the person to the assigned agent. Lofty says it runs on ChatGPT and Claude. The product page is Lofty AI Sales Agent, and the setup and price details I am using are in Getting Started with AI Sales Agent.

It skips hidden leads unless that agent has their own subscription, and it skips wrong numbers, do-not-call numbers, and landlines. An "AI identity disclosure" setting controls whether it introduces itself as AI. Lofty says that may be required where you live. The same help page starts at 200 leads for $60 a month. The quota does not reset monthly.

You need Lofty for this. It is not a bot you point at a different CRM or a spreadsheet. If the problem is the list itself, not the texting, the real estate lead generation guide is the piece about building the list.

9. Otter

Otter transcribes meetings. From its product page, it can join Zoom, Google Meet, and Microsoft Teams, or record on a desktop without a bot in the call. It writes a summary, pulls action items, and can send notes to Salesforce or HubSpot. The page lists a free plan with automated summaries.

It is not a real estate product. Use it after a listing consult so the follow-up quotes what the seller said. Read the transcript before you send anything. It can miss a name or a number.

10. HouseCanary

HouseCanary sells property data, valuations, and CMAs to investors, lenders, and agents. The homepage says the dataset covers 136 million-plus properties, and that image recognition reads room type and condition from photos. They also point verified agents at a product they call CanaryAI, along with CMAs.

NAR says 59% of members already use some CMA or pricing tool. HouseCanary is one vendor in that slot. A CMA you send a seller should show the comps, not only a model number. For a feed you control, see the real estate data solutions guide and the Realtor.com scraper guide.

Special mentions

Follow Up Boss

Follow Up Boss is a CRM with AI in the lead record, not a separate chat. Its AI page says Smart Messages suggest texts and emails from lead activity, suggested tasks are automatic, and tags surface high-intent Zillow buyers and qualified sellers. Call and history summaries are marked in beta. AI has no extra charge, but many of those features depend on FUB Calling, which costs extra.

The pricing page lists monthly Grow at $69 per user. The annual Grow block on that page shows $58 per user, with Calling at $33 per user. Pro and Platform name smart summaries, smart messages, suggested tasks, and predictive lead prioritization. Use it when the leads already live there. It will not write the listing or stage the photo.

How to choose without paying for all 10

Start from the hour you lose. Listing remarks point to ChatGPT, Gemini, or Copilot. Empty rooms point to REimagineHome, with a staged label. A walkthrough points to Matterport. Leads that already sit in Follow Up Boss point there. New leads on a Lofty site point to Lofty's sales agent.

Thirty-six percent of NAR members spend $50 to $250 a month on all of their technology.

Two habits matter more than the brand. Read the output against the house you walked. Label anything the client did not see in person, including a staged photo and a text that came from software.

A custom tool or a campaign

These products stop where their connections stop. A custom tool is the option when the fact you need is on a county site, a portal, or a list you already own. A campaign is the other version of that job: one listing, the photos, and the follow-up, run as one piece instead of ten logins.

We build both at BigiByte: property data, a lead list, and either a small assistant or the campaign around it. To scope it, use the contact page and send the source, the fields, and whether you need a tool or a campaign. This is not a quote.