As a team of seven spanning multiple industries, continents and generations, everyone came in with different instincts about what to solve and what to build. Being the only person with experience in both technology and journalism — albeit the youngest in the group — I saw my first job as bridging those perspectives, finding the common threads and turning them into a focused direction.

Once we had aligned on a specific scope (the Challenge section above), the next decision was defining our user. I leaned on my teammates' collective newsroom experience, asking questions about how editorial and business staff actually work before we landed on a dual focus. Editors would use the tool to monitor emerging trends and assess how thoroughly The Atlantic had already covered a topic. Social and audience engagement teams would use it to identify archived content that had become relevant again and resurface it at the right moment.

From there, the driving question became how to surface what's trending in a way that's actually actionable. I researched several approaches, including Twitter sentiment tracking and real-time APIs from prediction markets, before making the case for Google Trends for its scale and ubiquity as a reflection of people's current interests. Our technical co-leads, Matt and Sérgio, paired it with the Infactory API, which housed The Atlantic's full archive, and together we all mapped the product flow: what the tool needed to surface, in what order and why. We also established the system's core criteria: the Infactory API would search across The Atlantic's full archive, while Google Trends data would refresh weekly, giving editors a consistent pulse on what interests were emerging.



After breaking for lunch with the full cohort of hackathon participants, we moved into development. Drawing on my design experience, I began guiding UX and feature decisions, which Sérgio would then implement. A significant part of this process was prompt engineering; since the team was building through vibe coding, getting the right outputs required carefully crafted prompts to ensure the AI-generated article pitches and content surfaced were genuinely useful and aligned with The Atlantic's editorial standards.

It was here that I also realized what we had built wasn't particularly novel. Newsroom staff could already search trending topics manually without our tool. Automation alone wasn't enough, so I proposed we add more value by layering in a timeline of past coverage for each trending topic, alongside AI-generated article pitches that editors could assign to staff based on gaps in previous coverage and what was relevant today. That pivot became a core component of the product.

Toward the end of the day, the experienced journalists on our team called out a critical adoption challenge: it is notoriously difficult to get reporters and editors to adopt new tools, and we learned that Atlantic staff members are deeply embedded in Slack. So Matt prototyped a Slackbot extension that would deliver the same trending insights and article pitch recommendations directly into Slack, removing any friction between the tool and the people who needed it most.

Our team closed by presenting to the full hackathon, including The Atlantic's CEO and senior product leadership, and the Infactory CEO and board members.

An editorial intelligence tool that monitors emerging topics in real time, surfaces relevant archived content, generates a timeline of past coverage, and produces AI-driven article pitches.