What started out as a video marketing project quickly evolved into a video-creation tool for real estate thanks to the RESO Data Dictionary and Web API.
Nathan Brannen, Chief Product Officer at Restb.ai, and Paul Hethmon, CTO at Amplify Systems (formerly AMP Systems) collaborated on the project using RESO standards to run listing data and photo media through AI. | WATCH VIDEO (10:55)
Standards allow multiple organizations to provide a simple set of information, like an address or number of beds and baths, in a format that is readable across the industry.
At this point, Data Dictionary fields have the basics covered. But when looking at photos, finding the best way to express that information and how to standardize it becomes more of a dilemma.
A Variety of Variables
The challenge is that there are many ways to describe interior and exterior elements of a property or even the type of property itself. And even if you manage to label all of the elements in one property in a common format, there are hundreds of other ways to do so in every other property.
Consider a sweet and carbonated brown beverage in a can. Is it a soda? Is it a pop? Or is it a “Coke,” no matter the brand?
It turns out that everyone is correct, depending on where you are from. The same is true for real estate:
- Porch, Deck or Lanai?
- Rambler vs. Ranch
- Bungalow (U.S.) vs. Bungalow (Canada; which is a Ranch)
- Stove vs. Range Top
- Tap vs. Spigot
Determining what terms should be used presents issues for computer vision companies instituting AI technology.
Agents will use market-specific phrases to describe property details. That complexity has to be turned into something usable at scale.
- How do you decide what to show?
- How do you retrieve the details you want to use?
- How much customization is required between markets?
- How much work is required by the agent?
- How can we leverage AI?
In most cases today, arduous custom data mapping is necessary when working with clients. But the end user, which is the agent, doesn’t care about any of that. They just want something that works.
Standards to the Rescue
Standards help add structure to data in a way that is easy to use and can traverse different geographies.
Leveraging AI technology, products can turn visuals into adaptable data that works across markets.
By combining AI with standards, a list of specific details can be created and used across different properties and markets and, most important for this exercise, for marketing.
Some marketing prompts that are well supported in the Data Dictionary include:
- Architectural Style
- Room/Interior Descriptions
- Room Sizes
- Location
- Lot Size
When fed into an AI engine, photo elements can be broken down into a text format that can be carried through to a text-to-speech audio file.
And just like that, you have script elements for a video of photos stitched together over background music in about 15 minutes or less.
The Very Near Future
As the Data Dictionary advances, there will be even more opportunities to use standards for marketing purposes. For example, AI could automatically determine which of 50 bathroom photos is the best one to include in a listing presentation. It is already possible to look beyond what the image is to a more structured or ranked approach.
Amazingly, Brannen and Hethmon created a proof of concept for the video project in about a day thanks to Data Dictionary-compliant data accessible through Web API.
Even with just a listing ID in place, the system can do all of this.
With Web API Add/Edit, the new video can be sent back into the MLS system as a media element attached to that property.
“We’ve done things in other international markets where there aren’t standards to make the things the same across different areas,” said Brannen. “We have a great appreciation for how easy it was to work on this project because of the Web API and because of the standards.”
According to Hethmon and Brannen, innovation is much easier with standards in place.
“All of these standards help make things just work,” said Brannan.
