Sandy Springs outlines data strategy, aims to centralize city data and pilot AI projects

Keith McMillan, director of data strategy and analytics and AI integration, told the Sandy Springs City Council on May (work session) that the city’s data program is in an early stage but moving quickly toward centralized data and pilot AI projects.

“We don’t have a data problem as a city. We have a data integration problem,” McMillan said, describing the city’s current reliance on manual, descriptive analytics and the goal of moving to predictive and prescriptive analytics by improving data governance, implementing data transformation technology and creating a dedicated data team.

The presentation listed existing and planned uses of analytics and AI: an image‑recognition model used for the city’s canopy study; CrowdStrike for network security that uses predictive analytics; expanded use of Microsoft Power BI in the fire department for forecasting and natural‑language features; and an experiment with AI‑driven sentiment analysis for public comments on capital projects. McMillan said the public‑engagement work uses optical character recognition to extract PDF contents, cleans and combines data, then scores comments as positive, neutral or negative.

Nut graf: The briefing framed a push to move the city from manual reporting and departmental silos into a single, documented data platform so city departments can share consistent, higher‑quality data and test AI tools. McMillan said staff has completed about 15 departmental interviews and scored the city’s data maturity at roughly 1.25–1.5 on commonly used models.

Council members pressed on safeguards and scope. Council member John Paulson asked whether the city is capturing “enough information” to get reliable machine outputs. McMillan replied that the city is collecting large volumes of data but needs to model, cleanse and integrate it so departments can “see everything together.”

Several councilors raised data ethics and policy questions. Council member Andy Baumann asked whether predictive policing would be part of the effort; McMillan said police are “plugged in” through a technical working group but that predictive policing remains a longer‑term, ethically sensitive issue. Mayor Rusty Paul and others emphasized that elected officials and staff must set “guardrails” and said policy decisions will be required before the city advances analytics that affect people.

McMillan outlined next steps and near‑term projects: a request for data transformation technology to merge city systems; a data and AI literacy program with a dry run for power users scheduled for May 30; and the new Digital Innovation Initiative to coordinate tools across communications, IT and departments. He said the city is pursuing external partnerships, including an application submitted to the Partnership for Inclusive Innovation (PIN) that would pair Sandy Springs with a Georgia Tech team to test AI‑assisted reviews of permitting documents if the grant is awarded.

The city is also partnering with What Works Cities/Results for America on a certification process that assesses eight practices and about 40 subcriteria; McMillan said the city completed the assessment and expects a review decision in July.

Council members and the mayor praised the staffing and the rapid progress while repeatedly asking how and when policy questions would be elevated. McMillan said staff will classify data by sensitivity, develop governance documents, and return to the council with recommended policies and thresholds for when elected‑level decisions are required.

Ending: Council members did not take any formal action during the presentation; the briefing closed with direction to continue the work plan and report back on policy and classification recommendations.