Richmond AI Tool Missed Performance Goals
The city set a January 2027 deadline to improve the system's ability to resolve nonemergency calls without human intervention.
Updated on Oct. 14, 2026 in Artificial Intelligence
Live Poll
Do you trust AI-driven automated systems to handle your local government nonemergency requests?
Richmond has mandated that its AI-powered assistant increase its call resolution rate after the system failed to reduce the load on 911 dispatchers. Launched in April 2025, the tool has struggled to keep nonemergency inquiries away from live emergency staff.
Why it matters
The initiative represents a push to automate municipal services and alleviate staffing pressure on 911 centers. Officials must now determine if the current software can be optimized or if it requires a total replacement to meet public service standards.
The city spent $52,000 on the AI system, which is now expected to handle at least 25% of calls by January 2027. The long-term performance goal requires the tool to resolve 60% of all incoming nonemergency requests.
The players
Richmond
The municipal government overseeing the implementation of automated dispatch and inquiry software for its residents.
The details
The AI assistant operates by retrieving database information to address caller inquiries regarding RVA 311, animal control, and police information. When the system lacks sufficient confidence in its information retrieval process, it triggers a transfer to a live dispatcher. This mechanism failed to keep 81% of the 111,000 recorded calls from reaching the 911 center during the last fiscal year.
Timeline
The AI system launched in April 2025.
During the last fiscal year, 81 percent of calls were transferred to 911.
The city disclosed a new 50 to 60 percent resolution goal in September 2026.
The deadline for system performance improvements is January 1, 2027.
The Tech Race
The project follows a pattern set by the 911 emergency response service mandate, which necessitates strict accuracy in call triage. Richmond is currently balancing this requirement against the broader municipal trend of adopting LLM-based customer service automation.
Residents using RVA 311 or police nonemergency lines will interact with the current AI system as they call, though the majority will likely reach a human agent. The city’s decision to re-evaluate the vendor by January 2027 will determine if future call experiences become more automated.
The takeaway
The city’s shift toward automation highlights the difficulty of integrating conversational AI into critical public service infrastructure. Residents should track the January 2027 performance review to see if the city commits to the current platform or pivots to a new service provider.
What happens next
The city has established a January 2027 deadline for the AI to handle at least 25 percent of calls. If performance does not meet this metric by that date, Richmond officials will initiate the process of evaluating a potential system replacement.
Further reading
For more on how local governments are deploying large language models, visit the Artificial Intelligence section.
Source note: This article includes information reported by Axios.
Live Poll
Do you trust AI-driven automated systems to handle your local government nonemergency requests?
