Notice: Function Elementor\Core\Experiments\Manager::warn_removed_or_hidden_dependency was called incorrectly. Feature mega-menu depends on hidden experiment nested-elements. Please see Debugging in WordPress for more information. (This message was added in version 4.3.2.) in /home/smarter1/public_html/old.smarter.codes/wp-includes/functions.php on line 6260
Robotic Process Automation - Smarter.codes Skip to main content

Smarter.codes

Menu

Real Estate RPA. Assisting users to fill out long forms

Problem

Our client is a real estate agency with 100+ agents. For every deal, they must fill out 10+ lengthy forms in their back-office software. This was time-consuming for busy agents. And even impossible for newbie agents. Many would abandon forms in the middle, or would fill them incorrectly or slowly! The aim is to increase the number of forms that are filled in, correctly.

Solution

On one side we would keep open the back office software, and on another side, we would keep open our chatbot. Using plain English, our chatbot would prompt the user with questions, & automatically fill the forms based on answers. Users could observe forms getting filled out on their screens. The chatbot would navigate across the sophisticated sequence of steps – by navigating between screens of software – without the user having to leave our chatbot. On a few certain confusing fields, the chatbot would also advise users on what to fill in those fields.

Result

In 8 weeks, the team at SmarterCodes built an RPA + Chatbot, which our real estate agency client began to beta test across their team.

Approach
This RPA was built differently. Typical RPAs use the “selector” technique to identify GUI elements and operate on them. Ours rather employed Computer Vision to understand the GUI elements. This resulted in a better developer experience because some of the GUI elements in back office software weren’t operable with the selector technique. But with Computer Vision we were able to operate just like a human agent would.
To interpret the language of the user, we used a mix of Google Dialogflow and our own proprietary NLP AI. Google Dialogflow’s ready-made AI helped us prototype faster, and our proprietary NLP AI helped us interpret some curve-ball natural language inputs where