Citibot, a startup providing AI-powered municipal communication services.

Created a multi-channel chat platform that can integrate with municipal 311 system (as well as other city systems), a type of citizens’ hotline, allowing an AI-powered bot to field inbound inquiries from residents. The generative AI-based system—which uses a version of the model behind ChatGPT—can handle thousands of different types of questions from city residents. The system uses information sourced from city websites, documents, and frequently asked question pages, which co-founder and CEO Bratton Riley says helps guarantee correct answers.

Citibot’s AI system integrates directly with the back end of a municipal customer service system and becomes the liaison between a resident and the city department handling their 311 request, which often involves queries for information from the city or a notification of something that needs fixing. Depending on the city, the system can respond to chats via the municipal website or to direct texts from the resident. When a resident reports a pothole, for instance, Citibot routes the report to the proper department to handle it. When the concern is addressed, Citibot lets the resident who reported it know the issue has been resolved. If a resident asks a question the system doesn’t have content to answer, Citibot can learn from that information so it can answer similar questions in the future.

The system can easily link with the many different software systems cities use to coordinate their diverse work. Riley founded the company in 2016, and so far, Riley says Citibot has integrated into 14 different types of software systems used by municipal governments and is in use by 75 cities.

About 70% of 311 contacts from residents are a relatively easy fix—most people ask a simple question, report a small problem for the city to fix such as a cracked sidewalk, or notify the city of a lost dog. By tackling the simple stuff, Citibot frees up time for city staff to connect with residents who have more complex questions or concerns. “We want to help them drive those efficiencies, so that we can enable them to serve more residents,” Riley tells MIT Horizon.

Citibot started working with New Orleans in 2021, after the city saw a more than 300% increase in 311 inquiries during the COVID-19 pandemic. More than 180 people work fielding inquiries to New Orleans 311 and 911, but the influx of communications stretched staff thin. Now, Citibot handles nearly 2,000 monthly communications for the city. “What we’re able to do is help them serve more people and save a bunch of time in serving more people—and save the residents a bunch of time in waiting to hear back from the government,” Riley says. Aurora, Colorado also started using Citibot in 2021. The chatbot now answers 46% of resident questions automatically, leaving staff a lighter load.

Generative AI is able to use existing data to create brand-new content, like answers to resident questions based on information that’s already spread throughout a government website. When used in a chatbot, AI can help identify relevant information and get it to the user faster. “There’s all different kinds of things that AI can do,” Riley says. “We are using it for our mission, which is to create better engagement experiences for residents of all walks of life so they trust their government more.” For more on how AI can create new data, see How Generative AI Works.