Local news organizations across the country are wrestling with the same challenge: how do you cover more communities when resources are stretched thin?
At Connecticut Public, the answer wasn’t to replace journalists. It was to build tools that help journalists discover stories they might otherwise miss.
That’s the challenge Connecticut Public set out to solve with its Public Meeting Monitor, an AI-powered system that reviews government meetings, identifies potentially newsworthy developments, and alerts journalists to stories that may deserve further reporting.
The project recently earned first place for Most Innovative Use of AI in a News Organization at the 2025 Local Media Association Digital Innovation Awards.
On a recent episode of Keep It Local, Connecticut Public Investigative Editor Jim Haddadin and Senior Director of Data and Digital Services Susan Bell shared how the tool works, what they’ve learned during its first year, and why they view AI as a newsroom assistant — not a replacement for journalists.
The challenge: More meetings than any newsroom can watch
Like many local news organizations, Connecticut Public serves communities spread across a large geographic area.
Important stories often emerge from city council meetings, school board meetings, finance board discussions, and other local government gatherings. But monitoring every meeting in every community is nearly impossible.
Haddadin said the newsroom began exploring whether AI could help journalists become more aware of what was happening across Connecticut’s cities and towns.
The goal wasn’t to generate news stories. It was to help reporters identify conversations and developments that might otherwise go unnoticed.
How the Public Meeting Monitor works
The system currently focuses on government meetings published on YouTube.
When a participating municipality uploads a meeting, the Public Meeting Monitor automatically:
- Detects the new meeting
- Downloads the video
- Extracts and transcribes the audio
- Analyzes the discussion using newsroom-developed prompts
- Identifies up to three potentially newsworthy items
- Posts summaries into an internal Slack channel for reporters and editors
The summaries are intentionally brief. They provide enough information for journalists to decide whether a topic warrants further investigation.
“This is not a system that is designed to be producing content for an external audience,” Haddadin said.
Instead, the tool functions as an internal news discovery system.
Journalists still do the reporting
One point both Haddadin and Bell emphasized throughout the conversation was the importance of maintaining editorial control.
The Public Meeting Monitor does not publish stories. It does not write articles. And it does not bypass traditional reporting.
Reporters and editors review the information, verify facts, conduct interviews, gather additional context, and make all editorial decisions before any story reaches an audience.
Connecticut Public also provides guidance to newsroom staff on how the tool should be used and how information generated by the system should be handled internally.
The organization views the monitor as a source of story leads rather than a source of publishable journalism.
Stories that might have otherwise been missed
Over the past year, the Public Meeting Monitor has helped surface stories that went on to generate significant audience engagement.
One example involved a local finance board discussion that initially appeared routine but contained references to unusual spending decisions. The conversation prompted further reporting that ultimately led to a widely read story.
Another lead came from a Hartford City Council meeting where residents voiced concerns about a prominent nonprofit organization. The discussion alerted Connecticut Public to a larger issue that became the focus of extensive reporting.
According to Haddadin, two of Connecticut Public’s highest-engagement stories of the year originated from tips generated through the Public Meeting Monitor.
Just as importantly, the tool has helped reporters identify stories in communities that may not always receive regular media attention.
A second AI project focused on audience service
Connecticut Public’s AI experimentation extends beyond the newsroom.
The organization also developed Curio, a chatbot designed to help audiences find answers to common questions.
Bell said the original goal was to help audience support teams manage repetitive inquiries, particularly questions about PBS Passport, memberships, programming, and other frequently requested information.
Unlike the Public Meeting Monitor, Curio is audience-facing.
The chatbot relies on carefully selected information sources rather than searching the broader internet, helping Connecticut Public maintain greater control over the answers it provides.
Human support remains part of the process
Even as Connecticut Public explores AI-powered tools, Bell stressed that human assistance remains central to the organization’s approach.
If Curio cannot answer a question, users are directed to staff members who can provide additional help.
The same philosophy guides the Public Meeting Monitor.
AI may help organize information, summarize discussions, and surface potential story ideas, but journalists remain responsible for reporting, verification, context, and publication.
Lessons for local media organizations
One takeaway from Connecticut Public’s experience is that news organizations don’t necessarily need to build large, complex systems to begin experimenting with AI.
Haddadin noted that AI tools have become significantly more capable since Connecticut Public launched its first pilot in 2025.
Today, even a newsroom interested in monitoring a single city council or county commission meeting may have options for building workflows that help reporters track public meetings more efficiently.
For Connecticut Public, the focus now is on expanding adoption, improving the tools already in place, and continuing to explore practical applications that help journalists serve their communities.
The organization’s award-winning work offers an example of how AI can support local journalism — not by replacing reporters, but by helping them discover more stories worth telling.
***
Would you like to sponsor a future episode of Keep It Local? Our membership is comprised of more than 3,000 local media companies of every size and medium. We have a variety of ways to get you in front of key decision-makers within the industry.
Listen on Spotify: https://open.spotify.com/episode/2bIzgHv4Gc3joXmRTu07o2
Listen on Apple: https://podcasts.apple.com/us/podcast/how-connecticut-public-uses-ai-to-surface-stories-hidden/id1808196993?i=1000775809081
Watch on YouTube: https://www.youtube.com/watch?v=SRpPu8yy290
If you’ve listened to and enjoyed this podcast, please take a moment to leave us a review. That helps platform algorithms surface the content so that it can benefit other newsrooms.
Editor’s note: Artificial Intelligence was used to transcribe and create an initial summary, and then edited by a human.