Program Overview
We developed AI chatbots in collaboration with four local newsrooms: Atlanta Civic Circle, Chapelboro, Henrico Citizen, and The News Reporter. These are small, community-focused organizations serving English-speaking audiences in the Southeastern United States.
We began with a 30-minute conversation with each newsroom to introduce the goals of the pilot, outline the development schedule, and explain how the chatbot system would work. We also asked about the challenges they faced in day-to-day operations and how an AI chatbot might help address those issues.
While we did not guide or impose any particular design approach, the use cases identified by the newsrooms generally fell into two categories: customer support and content discovery, with some chatbots blending both purposes. Each chatbot’s functionality was tailored to the specific needs and priorities of its newsroom.
For example, the News Reporter was interested in using a chatbot to manage common audience inquiries, enabling quicker responses outside of business hours and reducing the burden on staff who routinely handled repetitive questions.
Henrico Citizen aimed to use the chatbot to help readers access less obvious information on its website. This included navigating archived stories and understanding how to advertise with the newsroom. For their small team, the chatbot represented a scalable way to improve access to scattered information.
Chapelboro noted that their website’s search function could be more effective. They wanted the chatbot to supplement that by helping users locate previous reporting and answer general questions about the newsroom.
Atlantic Civic Circle (ACC) sought to enhance audience engagement with its political coverage, particularly around local elections. With a large volume of content to draw from, they needed a chatbot that could surface concise, targeted information to help readers better understand local civic issues.
Following initial consultations, we developed chatbot prototypes for each newsroom using Zapier, a workflow automation platform. Newsrooms contributed source materials including customer service guidelines, FAQs, and a selection of relevant web content. These materials formed the basis of each chatbot’s knowledge base. With prompt engineering techniques, we structured the logic for how each chatbot would respond to user inquiries.
Within a week of the initial consultations, we built demo versions of the chatbots and invited newsrooms to test the prototypes and share feedback on functionality, accuracy, and user experience. Over the following week, we made iterative adjustments to improve performance and better align with each newsroom’s goals. Once finalized, we provided the chatbot code along with documentation on how to embed the bot on the newsroom’s website. We also offered direct support throughout the integration and launch process. The entire process from initial consultation to final deployment took approximately three weeks, with some variation based on each newsroom’s availability.
For a more detailed technical walkthrough of our development process, please refer to our previous workshop on building AI chatbots.