How a Texas Boomtown Used AI to Reinvent City Government

Last March, Jesse Elizondo was driving to his job as Assistant City Manager for Kyle, Texas, when he spotted a pothole in the middle of the road. He pulled over, grabbed his phone, and opened the city’s new 311 mobile app to report the hazard. By the time he drove to work the next morning, the city had patched the hole. 

In the past, that kind of rapid response would have been unimaginable. 

“A resident of Kyle might have to call multiple numbers to find the right department, report the pothole, and trust the city would eventually send a crew out to fill it,” said Elizondo.

Now nearly all of Kyle’s public-facing services have been unified by an AI-driven 311 system powered by Agentforce. With just a few taps, residents of this rapidly growing Austin suburb can report public safety hazards, broken streetlights, sanitation issues, graffiti, water leaks, and more. They can get an estimate for how long it will take to resolve an issue, track the status of their reports, and view a map showing where other issues have been reported. 

Since Kyle’s agent-driven 311 platform debuted last March, residents have submitted more than 12,000 requests. Nearly 90% of users got their questions answered or issues resolved during their first call. An average service request is now handled in fewer than 2.5 days — nearly twice as quickly as city officials had expected. And it’s already saved more than it cost to implement the new system by reducing waste and inefficiencies. 

“The citizens of Kyle no longer have to go through five different humans and a long, drawn-out process in order to get things done,” Elizondo shared. “Seeing things like potholes getting fixed in real time is why working in municipal government is such a cool job. You get to build systems like this and watch it serve the public instantly.”

Step 1: Identify the problem 

When City Manager Bryan Langley took office in 2023, he inherited a city government running on manual processes and saddled with information silos. Every department had redundant call centers and work order systems that didn’t talk to each other. People at city council meetings were voicing frustration about their inability to get answers to simple questions, said Joshua Chronley, Assistant Director of Administrative Services.

“Residents would call the Parks Department and think, ‘This person works for the city of Kyle; they must know what’s going on in the Water Division or Administrative Services,'” he said. “But the employees didn’t, so people were passed around the phone tree. We needed to figure out where the breakdown was.”

Working with the city council, Langley and Elizondo put together a strategic plan for a new, AI-powered 311 system to simplify residents’ interactions and service requests. The city formed an executive leadership committee to look at its options. Officials knew they needed a single point of contact that citizens could call on 24/7 and a way to combine all the information squirreled away in department silos. They also needed a way to push service requests to each department’s work order system, track the progress of each request, and confirm when they were resolved.

“We needed a platform that was user-friendly and customizable with other software,” said Chronley. “It had to be able to integrate with all the different pieces of software we use and become the core tool we built our vision around.”

We needed a platform that was user-friendly and customizable with other software. It had to be able to integrate with all the different pieces of software we use and become the core tool we built our vision around.

Joshua Chronley, Assistant Director of Administrative Services

Step 2: Lay the technical foundation 

Once the committee settled on Salesforce and Agentforce, the city’s IT staffers evaluated everything from hardware and software platforms to network bandwidth and staffing to ensure they had the right internal resources lined up. They implemented a single sign-on system for user authentication and applied to the Federal Communications Commission for a 311 license. 

The IT team quickly realized it needed a centralized repository to store the various types of data the city relied upon, so it built a data lake to house the text documents, images, CSV files, spreadsheets, geographic information systems, and other raw data that would be used to train the agent. 

“There were some heavy lifts behind the scenes,” admitted Chronley, “but our IT division really took up the mantle and ran with it. They built this giant data repository to feed our Salesforce machine.” Kyle’s 311 system integrates AWS and Data Cloud, with AWS handling system integrations and Data Cloud managing Salesforce databases.

Step 3: Test and refine 

Preplanning took a little under two months, and full implementation took another three. An external partner built the system’s underlying architecture, and another Salesforce partner helped the city create and test the Kyle311 agent, accelerating deployment.

City employees then stress-tested the web and mobile apps extensively inside a secure sandbox, collaborating closely with the IT and Public Works departments to ensure communications were working as intended, said Chronley.

“We were able to sit with the individuals on the other side and make sure we were all doing things the same way,” he said. “That allowed us to test in one cohesive motion without any communication barriers.”

Step 4: Educate employees and the public

A key part of the project was creating an internal knowledge base for staff members to answer questions when citizens call, including full transcripts and agendas from every city council meeting. That allows staffers to query the knowledge base when residents ask questions outside their areas of expertise so they don’t have to pass the callers on to another department. 

With the knowledge base in place, Kyle was able to reassign some call center employees to other departments in the city that needed help. But it also meant the remaining call center employees had to learn a new way to work, said Langley. 

One unique challenge was persuading employees to embrace the new technology and the idea of working alongside AI agents. 

“When we first started talking about the new 311 system, a lot of the staff were a bit skeptical. They worried they wouldn’t be able to keep up with the volume of calls,” he said. “We had to show both our policymakers and our employees exactly how routing all resident inquiries through a unified platform would improve services, save money, and result in better outcomes for our residents.”

Another challenge was making sure the public was aware of Kyle311 and knew how to use it. So the city launched a full public relations campaign, posting interactive videos of the app on the city website and even held a launch party when the platform officially went live on March 11 – appropriately, 3/11. 

Step 5: Fine-tune the process

Even after launch, the job wasn’t quite done. A few weeks after the app went live, it was clear a few minor tweaks were needed, added Chronley. 

In some cases, they had to change how 311 reports were being routed internally; for example, vehicles abandoned by the side of the road need to be handled differently than a junked car sitting on someone’s lawn. In other cases, it was about managing user expectations. Some streets in the area are maintained by the Texas Department of Transportation, not the city; the app needed to let citizens know potholes on those roads could take longer to fill.

To achieve this, Kyle fine-tuned the agent and expanded the knowledge base, aligning with internal structural adjustments.

Creating a blueprint for smart cities 

With the successful launch of Kyle311, city officials now hope to implement new features — adding geotags to each city asset, for instance, to allow for more granular reporting. Instead of simply noting that one of the lamps on a street is burned out, residents would be able to specify which one, further cutting response times. 

The city is also looking at incorporating data from traffic management systems into the app and using AI agents to simplify how residents apply for permits and inspections, Langley said. 

Kyle is not unique, stressed Chronley. The power of a platform like Agentforce makes it possible for any city to replicate what Kyle has done with its 311 system, but it requires true commitment and leadership from the top of the organization down. 

“You’ve got to get a handle on your data first, and you’ve got to make sure your systems all talk to each other,” he added. “Otherwise, you’re just creating more silos, and it’s not going to work.”

Cities need to start by envisioning how they want to serve their citizens, and work backward from those objectives, suggested Langley.

“You need to begin with the end in mind,” he said. “We always knew the outcomes we wanted to deliver to our residents. That helped us decide what type of product to offer, how to structure operations, and how to go about implementing it.”

Go deeper:

• Read why AI is pushing commerce to a major inflection point
• Uncover more data about agentic AI’s impact in retail 
• See how Eva, Engine’s agentic concierge, is setting the table for a revolution in hospitality
• Learn about agentic AI’s role in streamlining vacation planning