How Do Real Estate Agents Create Marketing Graphics With AI?

So you open a design tool, pick a template that almost fits, and guess at a color until it's close enough to call your brand's. Export, upload, repeat for the next listing, and repeat again after that. That's the treadmill most agents are still on, and it just stopped.

In this Live Training, Nick Krem, CEO and co-founder of the Krem Institute of Artificial Intelligence, and Eric Post, founder of Halo, walk through two new features built directly into Halo's chat agent: AI image generation and live web search.

1. How Do You Create On-Brand Listing Graphics Without Canva?

Halo already knows the style of your existing sites, the colors, the fonts, the layout of your property pages. When you ask it to create something new, a flyer, an open house graphic, a social post, it builds using that same style automatically.

One example from the training: a graphic made for an open house matched the exact colors and fonts of that property's own listing page, with nothing more than a plain description of what was needed.

2. Can AI Build a Social Media Carousel From One Text?

The clearest example of how fast this works is a full four-card Instagram carousel, built from a single text.

The entire prompt was: "I want to make a four-card carousel for this listing." Halo pulled the listing's images, matched the color palette and font, added the address, and delivered finished slides in minutes, ready to download and post.

That speed matters more than it might seem. Research on real estate marketing found that viewers spend 60% of their time looking at images and only 20% reading the listing description, and homes marketed with professionally edited photos see a 1,200% jump in social media shares. The visuals aren't decoration, they're most of the pitch.

3. What Else Can AI Image Editing Do?

The carousel is one example, but it's part of a much bigger shift already underway. The 2025 NAR Technology Survey found that 46% of REALTORS® already use AI-generated content, including listing descriptions, and that number keeps climbing. Image editing is simply the next piece of that same shift.

Beyond graphics built from scratch, the same tool handles smaller edits too:

• Remove a background for a clean, transparent cutout
• Turn a casual selfie into a professional-looking headshot
• Add virtual staging to an empty room
• Crop, rotate, or resize an existing image

Halo automatically decides which AI model to use based on what's actually being asked, so there's no need to know which tool fits which job. You just describe the result you want.

Graphics and edits solve half the equation. The other half is making sure what you're saying is actually accurate, which is where the second new feature comes in.

4. How Does Live Web Search Work Inside Halo?

See how this works →

Web search used to run only in the background, helping Halo build sites and research markets behind the scenes. It's now built directly into the chat agent itself, so you can ask it questions in the moment.

A few things make it work better:

• Use specific names or zip codes instead of general terms
• Share a direct URL if there's a specific page you want it to read
• Keep questions focused rather than broad
• Ask it to cite its sources

Common ways agents are already using it: prepping for an open house, researching a buyer, checking market positioning, relocation research, vendor research, tracking local news, researching nearby businesses, and even checking what other agents are doing to compete against them.

Search alone is useful. Paired with image generation, it gets a lot more interesting.

5. How Do You Combine Search and Image Generation?

The two features become more useful together than either one alone. The pattern is simple: research first, generate second.

One version: ask Halo to look up the schools, restaurants, and parks near a listing, then ask it to build a graphic highlighting what it found. Another version: ask it to find nearby local businesses and amenities, then create content featuring them.

Either way, the result stays accurate, because it's built from something Halo actually looked up, not guessed at.

Final Thoughts

None of this requires learning new software or switching tools. Every example here starts the same way: describing what you want in plain language and letting Halo handle the design and research work behind it.

If you're not sure where to start, try the carousel. Pick one active listing, and send the exact prompt used in the training: "I want to make a four-card carousel for this listing." See what comes back.