evidoria

← Back to browse

Good practice Imported

Sunny — Denver's Generative-AI 311 Chatbot Reaches 12% of Citizen Inquiries and Spawns a Copycat in Aurora

United States of America · Denver · See the United States of America profile · See the Denver profile

Evidence: Descriptive / self-reported Top 22% 73/100 · Ask Evidence Copilot about this practice

Denver's generative-AI 311 chatbot "Sunny" was handling 12% of all citizen-service inquiries by June 2024, two months after launch. Neighbouring Aurora copied the approach and Denver has since cut live-agent 311 phone hours -- but no independent impact audit has been published.

12%
Share of 311 engagements handled by Sunny (by June 2024)

Details

Maturity
Scaling
Promoter
City and County of Denver, Technology Services (311), with vendor Citibot
Period
2024–2025
Keywords
local government, citizen services, 311, digital government

Context

The City and County of Denver's 311 citizen-service line needed round-the-clock, multilingual coverage of resident inquiries.

Objectives

Deploy a generative-AI chatbot to answer 311 inquiries automatically, using the city's own website and knowledge base, with hand-off to a human agent for unresolved questions.

Activities

Denver launched "Sunny", built with vendor Citibot, in April 2024. Sunny draws answers from denvergov.org and Denver's 311 knowledge base and responds in dozens of languages. Denver has since reduced live-agent 311 phone hours.

Results

By June 2024, Sunny accounted for 12% of all 311 engagements, with usage still climbing into its busiest months (March-April 2025). The approach was copied by neighbouring Aurora ("Access Aurora") on the same technology, and Boulder and Broomfield have explored adapting it.

Conclusions

Sunny reached meaningful usage share and prompted documented replication by another city, but Denver has not published a rigorous, independently audited impact evaluation of the tool.

Implementation
Indicative cost
Low (< €50k) — Vendor-provided SaaS chatbot (Citibot) rather than custom-built infrastructure, consistent with a low-cost deployment relative to building a bespoke NLP system.
Time to results
Short (< 1 year) — Launched and scaled within about two months (April to June 2024) to handle 12% of inquiries, with continued growth through 2025.
Staffing & skills
City and County of Denver Technology Services / 311 team, vendor Citibot

Conditions for success

  • Chatbot grounded in the city's own published content and 311 knowledge base
  • Clear hand-off path from bot to human agent for unresolved questions
  • Multilingual support to match resident language needs

Common failure modes

  • No independently audited impact or satisfaction study has been published, despite the city reducing live-agent phone hours in reliance on the tool
  • No dedicated AI oversight or accountability mechanism for the chatbot is documented

Where it fits

Governance type
municipal government
Scale
single city (with replication in a neighbouring city)
Income level
high-income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful