Sentrics gives seniors a handsfree connection to their community

Amazon services support rapid customization

AWS Lex and several other AWS services allowed the company to create a chatbot device that connected to seniors’ televisions where it could provide an array of functionality. For example, seniors could view a community menu of activities and know which of their friends was going to play bingo or a swimming class, when they could join the next community shopping trip and what the dinner options would be in the dining hall. And they could do it all with voice technology for greater ease and flexibility.

“AWS Lex was a revelation,” said Patterson. “It brought us the capabilities we needed in a much more streamlined format, plus added capabilities that we didn’t know we needed. For example, the architecture made it very easy for us to provide seniors with lists of television shows and movie titles they could view on their TVs. We didn’t know that technology existed before Rackspace introduced us to it.”

Another AWS technology that supported the project was Amazon API Gateway, which is scalable and supports large data payloads, like audio. Amazon S3 provided storage for large audio files. Amazon CloudWatch Logs provided a single pane of glass for log views. AWS Lambda allowed the team to run code without administration. With AWS CodeCommit, the team could collaborate on code in a secure and highly scalable ecosystem. AWS CloudFormation supported the automated deployment of the application.

“Rackspace really understands AWS technology. They’re experts on the technology,” said Patterson. “I’ve been using AWS for 15 years, but I’m still overwhelmed by just trying to keep up with the new technologies and services AWS is continually releasing. After our first conversation, it was abundantly clear that Rackspace was going to be our partner.”

Learning how to train AI chatbots

Central to Sentrics’ new application was voice technology. Because the application was for seniors, many of whom have mobility challenges, voice activation was the key to ensuring accessibility for all. However, this also added to the complexity of the solution, because adding voice-driven functionality requires specific expertise.

AWS Lex provides the power of chatbot technology, which is powered by conversational artificial intelligence and machine learning. Chatbots are commonly used in “Help” applications on websites and in physical devices, like Amazon Echo. They first use automatic speech recognition to convert the sound of a human voice into words, and then use natural language processing to interpret the words into an instruction.

A chatbot must be programmed to recognize the “intent” of an instruction, as well as when it should expect any relevant “values” to take action. For example, if a person says, “I’d like a table for two at six o’clock,” the intent is to make a restaurant reservation, and the values are “two people” and “six o’clock.”

While training Sentrics’ chatbot to recognize potential intents and values, the team found that entering a large number of similar potential instructions was not as valuable as identifying distinct instruction patterns. After a few key inputs, the machine learning component kicks in. It begins to automatically make connections between intents and values as it learns to respond to a variety of voice commands.

Also, natural language processing can distinguish between concepts like today and tomorrow, correctly assigning the right date as the value for the instruction. This enables more natural conversation with the chatbot. The team was also able to leverage certain pre-existing value categories already developed by AWS, such as dates, numbers and even television channel names. This simplified instruction programming and accelerated development.

Racing to deliver to seniors in lockdown

To meet its intensified pandemic-driven deadline and bring its Engage360 application to seniors as soon as possible so they could engage virtually, Sentrics divided its project development into four key stages. At the outset of the partnership, the company asked the Rackspace team to complete all of the phase one tasks and potentially some of phase two items — all within six weeks.

Very quickly Rackspace finished all phase one tasks and was making headway on phase two items. Before the six weeks were up, Rackspace finished all phase two and three tasks and most of phase four items. At the same time, the team also built the API and the chatbot that sat behind the API.

Helping keep the team on task was Rackspace's own agile and scrum methodologies, innovative project management capabilities and engagement delivery framework. “Rackspace has a tremendous team, including developers, data scientists and a scrum master,” said Patterson. “They hit their deadlines and supplied us with accurate reports. The entire process was so stress-free. It was a continually affirming relationship.

Patterson added, “We were very impressed with their project management approach and engagement delivery framework. In fact, we’d like to adapt their processes into our own approach so we can gain more efficiency.”

The next step for Sentrics is to launch its current products in the senior marketplace and see what people say. “Then we’ll take that feedback and continue developing products and building connections between our products so they all work seamlessly together,” explained Patterson. “We also want to add functionality that improves seniors’ interaction with their family members, as well. We’ll definitely be looking to Rackspace's expertise to help us find the best ways to build and enhance our products and make those connections.”