Newsletter

Context Window: How AI and metadata power Chicago.com

By Sonali Verma

November 18, 2026

Service journalism has historically been looked down upon by many journalists as writers prefer to think big thoughts and practice capital-J Journalism. But recently, I’ve been part of many conversations about the importance of news as a utility. 

Many news organizations are realizing that if we are genuinely useful, people will continue to come to us. Put differently, if we are to remain relevant, we need to question ourselves: What purpose do we serve for the reader? Which of their needs are we fulfilling? 

I was intrigued, then, to hear about a local news project under way that is squarely aimed at being useful for the people who access it. Even more interesting: It uses AI, an instrument of Big Tech, to effectively connect people with their communities and bring them together on a social media platform that is owned and run by independent nonprofit and civic organizations, rather than a Silicon Valley corporation.

AI, metadata unlock community connection for Chicago Public Media

“What if we can collectively build digital spaces that are constructive, not extractive?” Melissa Bell, CEO of Chicago Public Media, asked. Her team, in partnership with New_Public, took on the task of building a social media platform that is a true alternative to Big Tech.

The result was Chicago.com, a hyper-local community platform launched in October that has an events calendar, user-generated guides to local places, and a forum where neighbors can share information. 

Screenshot of Chicago.com website

Each community will have a paid steward: “Someone rooted in the neighborhood who facilitates conversation, helps establish and maintain community norms, surfaces useful local information, connects residents to civic and in-person opportunities, and helps shape the platform as it develops.” Chicago Public Media is also hiring a Community Stewards Lead to build and lead the stewardship program. 

Readers can post in the forums and also build their own lists, like “things to do in my neighborhood,” or “a day well spent on a spring day.” Reporters also offer how-to guides that help the community understand various processes, such as how to cast a ballot.

Readers also get a newsfeed that pulls together content from Chicago Public Media and participating local news outlets. The platform applies topic labels, such as arts and culture or education, and location information to help people find news about a particular subject in a particular part of the city.

More than 4,000 people joined waitlists for the first five communities that were part of the launch. 

How they did it:

The team focused on using metadata to make its newsfeed more useful and locally relevant, said Matt Watson, vice president of product at Chicago Public Media. 

“We’re really trying to make it hyper local so people can say, ‘I’m interested in news about this topic in this neighborhood,’” Matt said at the News Product Alliance Summit.

The team used Backfield, an open-source tool, for entity and location extraction developed by Chase Davis, technical lead for the Lenfest Institute AI Collaborative and Fellowship. The goal was to identify locations at a fine-grained level – for example, not just a neighborhood but a specific intersection. Matt said the system could also infer location from contextual clues, such as an alderman being associated with a particular district. 

The location structure included 77 officially recognized community areas and about 250 neighborhoods. 

What went into creating it:

  • Making content discoverable without adding more work at publication: Matt said modern publishing workflows are already complicated, so he was hesitant to ask that journalists do any extra tagging. Instead, the team used AI-based metadata to supplement content after publication.
  • Gathering content from multiple sources: The newsfeed draws on content from Chicago Public Media and local outlets. Some partners could provide full-text RSS feeds; others could not, so the team had to consider building site-specific scrapers.
  • Working with imperfect legacy data: Chicago Public Media had consolidated different content management systems, including systems inherited from past migrations, so the existing metadata and content were “messier” than expected, and much of the cleanup lies ahead.
  • Using AI for prototyping: The team started visualizing the project on paper but quickly moved to an AI prototype built on Cursor “and just spun up what we thought the guides product could look like. We spent maybe under an hour doing that, typing it into a chat and building it out, and showed it to real users the same day to get their feedback. That’s a feedback cycle that didn’t exist before,” said Sarah Darby, vice president of data and analytics at Chicago Public Media.

    “I think it unlocks the ability for anybody to experiment, and the cycle in which you learn and test really shrinks. That’s a huge change we’ve seen in our organization.”

What’s next: 

The team is planning to make the enhanced metadata useful beyond Chicago.com by bringing it into the CMS or into a data layer and sending it to Google Analytics. Newsroom and business teams can then use it to understand which content resonates with its audience and contributes to engagement or conversions. 

“We haven’t had a chance to do this yet, but that will really be some of the most impactful work that we can do in terms of helping inform the newsroom and connecting it with the business side,” Matt said. 

Eventually, each of the 77 community areas will have its own forum.

Chicago.com is hoping to fulfill a user need that platforms such as Facebook and Nextdoor have been unable to meet, Sarah told INMA’s Media and Tech AI Week conference.

“If you’ve been part of those groups, you’ve probably found they can be quite toxic. They often have people posting about petty crime in their neighborhood and perhaps making assumptions about people. They often go unmoderated,” she said. Chicago.com wanted a platform that made “good neighborliness” the top priority.

“It is easy to commoditize information, but what people are really looking for is community and trust, and those can be differentiators. That’s at the core of what this new site is meant to do.”

Tools you can use:

This week’s AI Fellow spotlight: Mark Chonofsky

As a Lenfest Institute AI Fellow at Chicago Public Media, Mark Chonofsky is exploring how AI can expand access to local news, improve archival audio, and surface journalism more effectively. We spoke with Mark about the lessons he’s learned so far, what excites him most about AI, and where he believes the industry is headed next.

Local News Solutions

The Lenfest Institute provides free tools and resources for local journalism leaders to develop sustainable strategies to serve their communities.

Find Your News Solution
news solution pattern