Every community has a language of its own. Most AI has never learned it.
Large language models are trained on the internet.
A person can spend decades living in St. Petersburg and still discover something new about the city every week.
The city exists in thousands of places at once. In commission meetings and neighborhood plans. In nonprofit reports and development proposals. In local news archives, public records, and community conversations. Together, they tell the story of how the city works, what it values, and where it is headed.
Much of that information never becomes part of the data used to train general-purpose AI systems.
The Civic Language Model is AICOE's long-term research program to build AI that understands a community through its own records, institutions, history, and civic knowledge.
The city exists in thousands of places at once. In commission meetings and neighborhood plans. In nonprofit reports and development proposals. In local news archives, public records, and community conversations. Together, they tell the story of how the city works, what it values, and where it is headed.
Much of that information never becomes part of the data used to train general-purpose AI systems.
The Civic Language Model is AICOE's long-term research program to build AI that understands a community through its own records, institutions, history, and civic knowledge.
The biome principle
Every ecosystem develops according to its own conditions, history, and relationships.
Communities do the same.
St. Petersburg has its own institutions, neighborhoods, civic priorities, development patterns, and cultural history. The forces that shape the city today were built over decades through thousands of decisions, events, and relationships.
A person who understands St. Petersburg understands more than facts about the city. They understand context. They know how pieces connect, why issues matter, and how past decisions continue to influence the present.
That accumulated civic knowledge is the foundation of a civic biome.
Communities do the same.
St. Petersburg has its own institutions, neighborhoods, civic priorities, development patterns, and cultural history. The forces that shape the city today were built over decades through thousands of decisions, events, and relationships.
A person who understands St. Petersburg understands more than facts about the city. They understand context. They know how pieces connect, why issues matter, and how past decisions continue to influence the present.
That accumulated civic knowledge is the foundation of a civic biome.
What the CLM trains on
The Civic Language Model draws from a wide range of local information sources, including government records, public meeting transcripts, planning documents, local news archives, demographic data, economic data, and community-generated activity within Cityverse.
What it enables
The Civic Language Model helps AI systems understand local context.It can identify patterns in civic participation, connect information across institutions and neighborhoods, and support tools that help residents, organizations, researchers, and public officials better understand their community.
Within Cityverse, the CLM powers DASH, the platform's civic AI. Questions are answered using local knowledge sources that reflect the community's own information and history.
Within Cityverse, the CLM powers DASH, the platform's civic AI. Questions are answered using local knowledge sources that reflect the community's own information and history.
A long-horizon program
The Civic Language Model is a multi-year research initiative.
AICOE is building the data infrastructure, training systems, evaluation frameworks, and community partnerships required to support a city-scale language model.Each generation of the model will improve as additional data becomes available and new capabilities are developed.
The goal is a long-term civic resource built from the knowledge, records, and lived experience of a community.
AICOE is building the data infrastructure, training systems, evaluation frameworks, and community partnerships required to support a city-scale language model.Each generation of the model will improve as additional data becomes available and new capabilities are developed.
The goal is a long-term civic resource built from the knowledge, records, and lived experience of a community.
get aicoe news to your inbox
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
%2FNegative.svg){kind=small}
© 2026 AICOE - ALL RIGHTS RESERVED