Where did all these property generators come from?
Now that the AI services market is dense with property description generators, it’s difficult for real estate agents to choose the right one. To make the right choice, agents need to understand that these services can be categorized into three tiers.
Top tier: purpose-built, easy to use, accurate
Nila June‘s purpose-built, easy-to-use AI system pre-dates the GPT hype. Developed by professional writers in consultation with real estate agents, Nila June consistently produces high end descriptions that are accurate, creative, compelling–and compliant with fair housing regulations. Simple, single-purchase pricing aligns with an agent’s flow of listings.
Is there a downside? It can take an agent 7 or 8 minutes to complete the curated “property briefing survey” in order to tell Nila June everything about the home for sale. Considering the superior results, that’s a pretty minor downside. We doubt that most sellers would object to their listing agents spending a few minutes on Nila June in order to get the best possible description.
Mid-tier: purpose-built, easy to use, accurate
Generally built by technologists rather than by writers or agents, handful of services offer photo recognition technology and depend upon an address in order to identify a list of locational amenities. These tech-heavy services might be right for listing agents who have an interest in photo recognition technology, or who are unfamiliar with the local features surrounding their listing. Perhaps they are new to the area, or don’t have an opinion about whether to mention this or that park, shopping district, employment area, or school. We suspect that few agents have a yen for photo recognition QA, or are unfamiliar with the markets in which they operate.
The downside here is pretty significant. These systems get their text output (AKA the final product!) from large language models, such as OpenAI’s GPT or other large language model (“LLM”). These LLMS are bedazzling but imperfect, as has been well documented. Their frequency-based, predictive nature leads to descriptions that are sometimes wildly inaccurate yet ordinary sounding, and in violation of fair housing language regulations.
Bottom-tier: simple overlays to GPT or other LLM
The vast majority of property description generators are simple overlays to GPT or to some other LLM.
Downsides are numerous. These services present all of the disadvantages of the mid-tier, without any of the possible advantages that might (for some agents) arise from the photo recognition tech, or from the feed of buildings, parks, or districts that might be worth a mention. Rather than choosing one of these bottom tier property description generators, agents would be better off going directly to https://chat.openai.com/ and working on their “prompt” skills.
Is there an alternative to Nila June
For agents who want to engage with photo recognition technology, or who are unfamiliar with their local markets, who who are interested in editing raw output from ChatGPT, there are numerous subscription-based services that can help.
For agents who want spend just a handful of minutes to get the best descriptions, there is no alternative to Nila June.