Getting in contact with local government can be a slow, frustrating process. What if you could could send a text message from your phone to seamlessly interface with your government to suggest changes that can make your life and your neighbors' lives easier?
Citibot’s mission is to fundamentally change how citizens contact and interact with their local governments through digital channels. Specifically, it wanted to develop a chatbot that allows for engaging, conversational experiences through voice and text that can be customized and used on mobile devices, messaging services, and popular chat platforms such as Facebook Messenger and Slack. The goal was to let people use existing messenger applications so they would not have to download another app or visit a website.
Citizens and local governments use the chatbot tool to drive civic discourse and change. Using text messages and messenger applications, citizens can report issues, ask direct questions and get instant answers, receive notifications, and send personalized messages, allowing governments to provide more effective communication and make information more easily available to citizens.
After launching the initial version of the Citibot chatbot, the organization engaged with AWS Partner Network (APN) Advanced Consulting Partner and AWS ML (Machine Learning) Competency Partner TensorIoT to improve its natural language processing and provide expertise with Amazon Lex. Using the same deep learning technologies that power Amazon Alexa, Amazon Lex is a fully managed service that developers can use to build conversational interfaces into any application.
Because Citibot’s original chatbot had a tendency to get stuck on certain logic points, the organization sought to make it easier for users to interface with governments and reduce incorrect response messages. They wanted to help governments more efficiently and effectively triage issues raised by citizens through the chatbot by improving the logic behind the chatbot and removing manual ticket review. By reducing manual review time, the chatbot could give time back to the customer service staff to focus on addressing more citizen requests.
The TensorIoT team worked closely with Citibot to give its chatbot some personality and evolve its tone. The team helped Citibot optimize its AWS architecture to integrate with Twilio, a developer platform for communications that enables companies to embed voice, video, messaging, and authentication directly into applications, with an Amazon Lightsail application serving as the frontend handler. The company uses an Amazon API Gateway as an interface for Amazon Lex. Amazon Lex tweaks responses and high-level conversation category building. If more powerful business logic is required, Lex invokes an AWS Lambda function. Amazon DynamoDB stores conversation history and unexpected phrases that can be used to improve the model and build additional models.
Citibot has increased citizen engagement with local government by 94% and has made local government 112% more efficient. Citibot has earned positive feedback since its chatbot updates went into production and is now in cities and counties across the country. The company continues to mature and refine its natural language processing with the help of TensorIoT.
Read more about how Citibot worked with TensorIoT to build a citizen-friendly chatbot experience.
TensorIoT was founded on the instinct that the majority of 'compute' is moving to the edge and all 'things' are becoming smarter. The company is creating solutions to simplify the way enterprises incorporate Intelligent Edge Computing devices and their data into their day-to day-operations.