Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Ai Data Strategy topic
No spam. Unsubscribe anytime.
Sandy Springs staff outline enterprise AI rollout, pilot shows productivity gains
Summary
City staff presented an enterprise AI and data strategy, reporting a small pilot that expanded from 20 to about 37 users, rollout of a licensed Gov AI platform for employees, an imminent public chatbot launch, and plans to modernize the city’s Lucidie service‑ticket system.
Keith McMillan presented the city’s data strategy and AI integration update, saying a controlled pilot that began late in 2025 grew from 20 to roughly 37 licensed users and demonstrated time savings and higher productivity. McMillan described the decision to license an enterprise AI (gov.ai) so the city controls training and connectors, and to use retrieval‑augmented generation (RAG) with folios that contain policy documents, Municode ordinances and other city content.
The presentation laid out several near‑term deployments: enterprise Gov AI for staff with introductory and advanced one‑hour training sessions, a public chatbot soft launch on the city website scheduled for the next day, and expanded AI support in permitting through OpenGov. McMillan said the pilot used Perplexity and that the enterprise license prevents city prompts from being used to train external models: "We chose Perplexity… We were in our own space," he said, adding that role‑based access will limit who can connect to sensitive content.
Why it matters: staff said the initiative is intended to reduce friction in service delivery, surface sourced answers about ordinances and policy, and free call‑center staff to handle complex requests. McMillan noted training and monitoring efforts: a first training session drew about 100 attendees, prompts jumped to 167 in one day, and surveys at 30/60/90 days measured use patterns and time saved. He also said retention and public‑records handling for AI chat threads remain unresolved and that a decision on thread retention was expected in October.
Council members asked for an implementation timeline and communications plan. Councilmember Melissa Mueller asked for a high‑level timeline by quarter and asked whether Lucid modernization will include resident self‑service; McMillan said selecting a Lucid replacement is imminent and that resident ticket visibility is a top priority. Councilmember Mary Ford and others pressed on records and retention; McMillan said governance and retention windows are under active discussion and that role‑based access and data detectors (to flag or redact phone numbers and SSNs) are in place.
The presentation also described planned open elements: a grants assistant web crawler to gather federal/state funding opportunities, and a clerk‑ready minutes tool the city developed (agentic AI) that downloads meeting transcripts and agenda items to assemble minutes. McMillan said that city‑developed code would be open‑sourced for broader municipal use and that some vendor integrations (Municode, CivicClerk) will be used to improve search and ordinance comparison features.
What’s next: staff signaled phased rollout this summer for many features, office hours for staff, continuing performance management and monitoring, and a pending retention policy decision due in the fall. Several council members urged an accompanying public communications and training plan for residents when services (chatbot and self‑service ticketing) become public.
