You are here: GeekEstate Blog » Development » Bots, Aren’t So Elementary, Watson

Bots, Aren’t So Elementary, Watson

Bot, Why?

Sherlock Holmes once said, “It is my business to know what other people don’t.”

Real estate agents can certainly attest to this as they are inundated with requests from prospective home buyers and sellers for information. This information is a real estate agent’s most valuable asset, but what does this mean for the consumer?

It means we have to wait 5 hours for a response with no real context…

iowa_realty_FBmessenger

 

Bots Are Better People

Bots are built by context and what consumers want is a simple, informative response in a timely manner. They don’t want to have to give up their valuable time or information to wait around for you to call/text them, or worse, make them join a drip campaign.

Real estate is a personable business and chatbots are software built for conversation. These artificially intelligent conversational interfaces live inside the messaging apps you use every day and they respond right away, unlike their human counterpart.

 

Screen Shot 2016-04-23 at 2.16.36 PMBots Aren’t So Elementary

At their core, bots are powered by some serious machine learning that agents, brokers and consumers will never have to see. This is the real beauty in bots, where all your technology can be leveraged through a simple conversation.

The data science team at Settln has built Holmes (hence all the Sherlock Holmes references), the real estate bot that answer’s all your real estate questions in real time.

To start building a bot, you send it to either one of two schools – Rules or Machine Learning. There’s a bad cold start problem when building bots, because even to build one with medium complexity takes a hefty training dataset of conversations.

To make sure Holmes graduates, we put it through both schools. It started at Rules School using manually crafted conversations like the one Drew wrote. Behind the scenes the engine is looking at the text corpus, which is just a bunch of structured text, in order to find context.

 

Screen Shot 2016-04-23 at 2.00.03 PM

Bots Structure Unstructured Data

Holmes found that the context of this conversation was about a school, a listing, and a showing time. This context was derived from N-gram models using the text corpus.

 

Bots Get To Know You

Bots are great companions. In fact, I talk to multiple bots everyday. The CNN bot knows what news I like, the Weather Channel bot knows where I live and the H&M bot knows my style of clothing.

Bots replace your drip campaigns with timely, relevant messages

Bots replace your drip campaigns with timely, relevant messages

Holmes also knows the properties you will like and that’s thanks to his best friend, a graph database. This database helps Holmes take the structured data it made from conversation and structure it for models that predict home buying behavior and interests.

Graph databases open up an almost infinite amount of doors for Holmes.

Bots Open Doors

Consumers are always knocking on your door, requesting information of all kinds. Bots open up your technology to all your clients at all times.

Retsly_product

For example, our predictive days on market statistical model predicts how long a property in Arizona will be on the market for. A potential home seller could simply converse with Holmes to get an estimate on how long their property would be on the market.

Holmes uses predictive models to answer seller's questions.

Holmes uses predictive models to answer seller’s questions.

Bots live where context and convenience matter most, which is what leads to conversion. By interfacing with the technology already used by consumers, brokers, agents, lenders, and builders, bots can build real relationships, in house.

With Holmes, it’s never a matter of moving to get what you need, it’s a matter of where you want to move to.

Bots Are In Beta

Questions arise in conversations and conversations live in the messaging apps we use daily. The future of bots will allow us to bypass apps and search engines all together and instead, make purchases or find valuable information with a simple text.

However, to get to an “appless” world takes bots much smarter than we see today. This technology is very raw still. To integrate bots into these platforms takes more red-tape than development and documentation is very scarce. So be leery when building, bots are in beta.

Consumers have access to unimaginable capabilities with bots and all with a simple text message.

About Nathan Joens

Nate is the co-founder of Structurely, building artificial intelligence for real estate to help personalize interactions buyers and sellers have through messaging. Nate works with a team of 3 data scientists specializing in artificial intelligence technology who love to solve the tough problems they are faced with in real estate data.

This entry was posted in Development and tagged . Bookmark the permalink.
  • Nathan, can you provide an example we can see in action?

    I would like to suggest you use a new example because any good Real Estate team is not going to make a lead wait 5 hours for a response and then just tell them to visit their website.

    A real example using a good team makes the client wait only about 5 minutes or less, then discusses the clients needs in depth while asking them important questions to get to know them better, and most importantly starts building rapport with the lead, so they become a client.

    If you turn this function over to a bot you have lost your chance to build rapport and you probably lost a client. That is the reality, and Realtors are not going to do anything where they might lose a very valuable client.

    I am not saying there are no use for bots in Real Estate, but perhaps a better example is needed. We have a feature where with one click the computer runs through hundreds of comparable searches and comes up with the best comparables. We do not call it a bot, but the function is similar. This saves the user time having to figure out what the best comparables are. Many times the computer is right on the money, sometimes some additional tweaking is needed.

  • I tried Holmes on the web and via text. Nothing was afoot….

2008 - 2017 GEEK ESTATE · ALL RIGHTS RESERVED - THEME BY Virtual Results
Hosted by Caffeine Interactive