The problem became clear about two years ago for the city of San Jose: Far too many city residents were calling 911 to complain about potholes, broken streetlights and abandoned vehicles, despite the presence of a 311 system for exactly those service requests. That was driving down 911 response times and the City Council wasn’t happy.
Now, through a partnership with Google.org and several tech vendors, the city’s 311 system has been reimagined into an omnichannel network that lets residents request service in several languages and diverts 30,000 calls a year from 911, Jerry Driessen, the city’s chief technology officer, told Techwire.
The 311 system includes a call center, virtual agents, chatbots, Web portals and online forms designed to allow residents to quickly request services. It also addresses digital equity, as Google translation services provide virtual agents to smoothly communicate with residents who speak Vietnamese and Spanish as well as English — important in one of the most linguistically diverse cities in the nation. The system also can pick up idioms, slang and differences in dialects.
Driessen said a key to the transformation was city Chief Information Officer Rob Lloyd persuading Google.org — the tech company’s charitable arm — to assign six Googlers to the project at no cost.
“They did a full analysis and recommended that we change our call trees and promote a more omnichannel approach, so from the residents’ perspective it doesn’t matter how I get into the system because the look and feel and service is similar,” he said.
“We grew our San Jose 311 app and are adding additional services and machine learning language translation,” Driessen said. “We can train the tool and it has 98 percent accuracy compared to about 68 percent with the free tools we had been using. And we added the virtual assistant to alleviate possible stress to callers.”
Besides Google Cloud and Google’s Contact Center AI program, the project includes Mission Critical Partners, a specialist in 911 and 311 systems; SpringML, which set up the virtual agent and machine learning language translation; Carahsoft, a government IT firm; AST, which built and maintains the code for the San Jose 311 app; Altigen, which provided the call center’s call handling system; and Twilio, which serves as the Communication-as-a-Platform service. Twillio essentially functions as a broker between Altigen and Google’s Dialogflow. The app is built in the Oracle service cloud, and it’s connected via the cloud to other city systems.
Driessen said the updated system cost under $500,000 and is receiving about 240,000 calls a year. The mobile app has delivered over 200,000 requests as well.
“The most important aspect of all of this is that it was focused on the users — the residents of San Jose,” Driessen said. “Human-centered design was initiated by Google.org, and our internal staff worked with our partners to execute it. Staff also performed extensive usability testing and worked with human translators to train the platform and optimize the quality of the machine learning language translation.”
The CTO said the city plans new services and support in additional languages, such as Mandarin, in the future. One new service, to automate requests to haul away large household junk items, will be unveiled in March in conjunction with the second annual National 311 day March 11. Virtual agents will support the program in English and Spanish and use GIS programming to identify which of the city’s four private trash haulers serves callers’ homes.
