Dive Brief:
- Salesforce this week launched Agentforce for Public Sector, a “digital workforce of intelligent AI agents” designed to assist government employees with clerical work.
- Local governments can build and customize the AI agents to handle tasks such as answering common questions, processing license renewals, spotting complaint trends to flag pain points and recruiting, according to Salesforce.
- In Kyle, Texas, which incorporated the AI tool with its 311 platform, Salesforce reported it answered questions or resolved problems for nearly 90% of users on the first call. “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,” Kyle Assistant City Manager Jesse Elizondo said in a statement.
Dive Insight:
Numerous companies are developing AI tools for the public sector. This year, OpenAI, one of the largest AI developers, launched ChatGPT Gov for government employees, and major U.S. cities have announced plans to incorporate AI into daily workflows.
Saving local government workers time is a selling point for the AI products. Nasi Jazayeri, Salesforce public sector executive vice president and general manager, said in a statement that Agentforce “will power a more responsive, agile, and effective government.”
A 2025 Granicus survey of 1,400 municipal employees found that 40% used AI, largely for research tasks (62%). Of those surveyed, 75% saw the benefit of AI growth in local government operations, but 66% “aren’t sure if AI will significantly impact government operations.”
One of the biggest challenges to adopting AI in the public sector is accuracy and trust, according to Joseph Farsakh, co-founder of Helios, an AI-native operating system for government.
“There's no room for AI hallucinations when determining someone's housing eligibility, calculating benefits, or making resource allocation decisions,” Farsakh said in an email. “You have to build systems where every output can be audited back to its source data, and where generating plausible-sounding but potentially incorrect responses is a no-go.”
State and local government IT leaders have expressed caution regarding AI’s implementation, including regulatory and security concerns. The Trump administration’s AI Action Plan calls for the removal of “onerous Federal regulations that hinder AI development and deployment.”
Farsakh said AI regulation isn’t sufficient, but at the same time, “we also don’t know what we have in order to regulate it.”
“We didn't regulate cars before the car was invented,” Farsakh said. “We're still in the innovation and building phase, discovering what these AI capabilities can actually do in the public policy space.”
