Selma city manager demonstrates Madison AI; council weighs $27,000 implementation and ongoing governance costs
Loading...
Summary
City Manager Jason Keane demonstrated Madison AI, a searchable repository of Selma’s resolutions, minutes and policy documents. Council asked about recurring licensing ($23,000 governance module) and guardrails to prevent access to privileged or closed‑session records; staff recommended internal policies and audit logging.
The Selma City Council on Nov. 18 watched a staff demonstration of Madison AI, an AI‑powered document search and briefing tool the city has been piloting. City Manager Jason Keane showed how the system can pull historical resolutions, meeting minutes and adopted policies and create chronological briefings for development projects that span years.
Keane said staff uploaded resolutions dating to the 1940s and that the system’s search returned a 1993 resolution governing parking fines as an example. He described features that will help staff draft staff reports, summarize voting histories and pull related citations. Keane emphasized that Madison AI is an internal, closed‑loop tool that uses only files the city inputs and requires a city login.
Council members raised questions about cost, data security and legal exposure. Keane and the city attorney, Costanza, said the purchase includes implementation and licensing and that an annual governance module would cost about $23,000; Keane said $27,000 covered initial implementation and training. City Attorney Costanza warned that confidential or privileged closed‑session documents must not be uploaded; he also flagged risk if employees could download and later disclose documents outside official channels. Staff said every query is logged in the system and that access is controlled by city email logins.
Council members and staff discussed policy guardrails, citation practices and audit trails. Some council members suggested requiring explicit notation when a staff report or resolution was drafted with AI assistance. Several speakers in the public and council praised the tool’s potential to reduce staff search time; one resident, Matthew Rodriguez, told the council he believed the system would pay for itself quickly by saving staff hours.
No final contract was adopted at the meeting. Council asked staff to return the Madison AI contract and detailed costs for review and to draft an AI governance policy and an updated social media policy that would work alongside it.
