Lewisville council hears AI strategy focused on governance, data controls and staff training

Aaron Gelza, the city’s presenter on artificial intelligence, told the Lewisville City Council on Jan. 7 that the city is preparing a formal AI framework that prioritizes responsible use, transparency and safety while focusing on business needs. "Our AI vision: we'll use AI to work smarter, deliver better services, predict needs from our data, and do so with a commitment to safety, transparency, and fairness," Gelza said, summarizing the committee's guiding language.

Gelza said the committee — led by the deputy city manager and including department directors and staff — mapped a staged approach: identify clear use cases, measure organizational maturity, and prioritize projects that are feasible and align with business goals. He repeatedly stressed buying commercial solutions rather than building large language models in house: "We're going to buy the majority of our solutions," he said, while noting occasional extensions of vendor models to tie them to city systems.

The presentation laid out six responsible principles and five maturity categories (governance, data management, people, process, technology). Gelza told council that Lewisville is between the "incorporation" and "proliferation" stages: several pilots and proofs of concept exist, but few projects are at scale. He said staff will bring a generative-AI policy to council this quarter and continue work on a data-classification directive already funded this year.

Council members asked about vendor platforms and data protections. One council member mentioned Microsoft Azure and Copilot and asked whether Lewisville would use private or public cloud services; Gelza said staff is evaluating Azure, Amazon and other offerings and noted the difference between private and public-hosted models. "If you can have everything in transit to your own organization, then you minimize your risk," Gelza said about private-hosted environments. He added the city has provisioned about 40 licenses for staff pilots of generative-AI tools.

Council members also pressed on where city data currently resides. Gelza said data is scattered across Laserfiche (on-prem), H and S drives (on‑prem), and Google Drive (end-user data), and that some specialized repositories (for HR or HIPAA-like data) are on separate systems. He told council staff are pursuing backup and monitoring tools for Google Drive and are working on data classification: "You can't protect what you don't know," Gelza said, noting classification is a foundational step to build further safeguards.

On workforce effects Gelza acknowledged upskilling will be necessary and recommended training and regular "ask me anything" sessions for staff as adoption grows. He recommended onboarding training that would include responsible-AI practices, similar to security awareness training.

The council did not take formal action on the presentation but directed staff to return with a generative-AI policy and further work on data management and procurement options. Gelza said the next steps include bringing the policy to council this quarter and continuing the committee’s work on maturity, tooling and prioritized pilots.

The presentation ran as part of the workshop session; council members followed with questions about vendor choice, data locality and whether limits on employee use of generative tools would be enforced by policy and training. Gelza said staff will proceed with an emphasis on aligning use cases to business goals, prioritizing low-risk integrations first, and strengthening governance and data controls.