Downtown Concord viewed from the rooftop of the "Love Your Neighbor" Building, which houses Gibson's Bookstore. Credit: CATHERINE McLAUGHLIN / Monitor

This project began with a walk down Main Street.

For a few hours a day over a total of about a week, reporters Catherine McLaughlin and Rachel Wachman catalogued each storefront and took notes: the exact address, its function, info about hours taped to the front door.

With the aim of capturing the hours and offerings of every downtown business, it became clear that boundaries for data collection needed to be set. No matter where the line is drawn, it’s imperfect: there’s no hard stop or stake in the ground that defines where downtown starts and ends or what counts as a storefront.

Using the entirety of Main Street as it appears on a map wouldn’t make sense. Stretching from Bouton Street to Water Street, South Main and North Main extend beyond the bounds of what most people consider the true downtown. Conversely, leaving out side streets like Warren and Pleasant would exclude mainstay businesses of the commercial neighborhood.

Decisions about where to draw those lines were driven by the project’s goal: to explore downtown as the walkable business and cultural center that it is.

Loosely, the map’s endpoints align with the boundaries of the 2016 Main Street Project, from Centre Street South to the Draft, including the first block of key feeder streets like Depot Street and extending to the Main Street side of Storrs Street. For the zoning nerds out there, the project boundaries almost exactly align with the city’s Central Business Performance District.

When it came to which organizations and businesses to include, some spaces with Main Street addresses were excluded that didn’t read as storefronts or have a real sidewalk presence, like upper-floor offices with just a lobby on Main Street or some older homes retrofitted into business spaces, like Waters Funeral Home.

Once the boundaries and list of businesses were set, exact geographic locations for each were determined using MyMaps, a Google Maps extension.

Property and business records, outreach and research sourced who owned each building, how long businesses had been downtown, their hours, who the most recent tenants of vacant storefronts were, and other information. 

Some of this data had to be generalized. For example, using only the listed box office hours for the Capitol Center for the Arts paints an incorrect picture of when that space is buzzing, and there’s no exact schedule for the days and times for shows and other community events there.

A few gaps emerged in the research and documentation, too, especially when it came to the exact tenure of a business. Many of the Main Street stalwarts have moved multiple times or changed owners or taken a new name.

A set of categories was used to typify each business or organization, designed to accurately characterize the mix while not creating so many buckets that the categorizations couldn’t be meaningfully summarized.

To help with the coding and features of the map, a complete dataset was uploaded into the Claude artificial intelligence platform, alongside a description of the envisioned interactive features.

On a custom mapping platform called Mapbox, Claude coded and built the map and its interface. Reporters tested what was drawn up and made adjustments and improvements.

The back and forth also required a great amount of vigilance for the integrity of the data. Frequently, Claude would wrongly apply, misplace or misread the information in visualizing it – a regular pitfall of AI platforms. Reporters meticulously confirmed not only the functionality but also the accuracy of what the platform drew up.

This is keeping with the Monitor’s policy of using artificial intelligence as a tool, but not allowing it to generate content.

The map represents a single moment – June 2026 – but will change as the dataset is updated. Additions, corrections and changes over time are easy and expected. 

To be clear: this digital map was structurally built and coded with artificial intelligence. The data, analysis and vision behind this project, however, was organized, driven, checked and executed by real, flesh-and-blood reporters. 

Feedback, questions, suggestions and any changes are welcome. Experimentation and play with the features are encouraged.