Portland – October 27, 2026 -- Multnomah County, Oregon, migrated a decade-old SQL Server environment to a modern data platform in five months, a timeline that public sector organizations often stretch over years, according to a session at this year's Snowflake Summit.
Multnomah County replaces undocumented SQL Server estate with AI-assisted migration
The county's legacy environment spanned dozens of business intelligence workloads and had been built by engineers who had since left. Documentation was sparse, the architecture was inconsistent and costs were climbing.
Using Snowflake's SnowConvert AI, the county combined deterministic code conversion with AI-assisted handling of dynamic SQL, code fixing and automated test case generation, replacing manual rewrites. The county reported a shorter migration timeline, lower future spend on legacy SQL Server licensing and a standardized architecture. That architecture now covers previously siloed lines of business, including health and human services, public safety and one of the largest library systems in the U.S.
County team urges starting migrations a year before license deadlines
Multnomah began migrating three months before one of its renewal agreements, which left too little runway. The team recommends starting a year or more ahead of any licensing deadline.
It also advises evaluating which workloads to keep, archive or abandon, which streamlines the new environment and cuts project cost. Code conversion may take weeks, while validation and parallel-environment testing take months. Senior Database Administrator Michael Hanna said time spent preparing data "pays off multifold in the end."
San Francisco targets a citywide unified data platform
The City and County of San Francisco is at an earlier stage. Chief Data Officer Soumya Kalra described departments that have historically run isolated systems, homegrown databases and separate definitions of "good data."
The stated goal is a citywide unified platform that ends the departmental silos blocking cross-agency insight. Before unifying data, Kalra's team must build trust, shared standards and internal champions. It is involving stakeholders beyond IT, such as analysts in the public health department. The city treats change management, rather than technology, as the factor that determines whether modernization lasts.
Snowflake executive ties AI readiness to data strategy
Federico Zofaly, Snowflake's Director of Migrations and Modernization Center of Excellence, said: "There is no AI strategy if you don't have a data strategy." Legacy systems create data silos that block AI adoption, and organizations that have not modernized cannot support AI workloads.
The sequence outlined at the Summit runs from modernizing the foundation to unifying the data estate and then activating AI workloads. Architecture decisions made now should not need rebuilding within three to five years.