How Stacker Built an AI Recommendation Engine Clients Actually Trust

Ken Romano, SVP of Product at Stacker, discusses how his team built successful AI workflows both internally and as product features. He shares the story of creating an internal data querying assistant that initially failed because users found a blank interface overwhelming, but succeeded after adding Slack integration and example questions. Romano then details the development of Sparks, Stacker's new AI-powered content recommendation tool that helps brands determine what stories to write. Built over 90 days using Claude with extensive prompt refinement, Sparks analyzes brand data and network performance to provide weekly story recommendations. The conversation covers practical lessons about AI adoption, including the importance of finding internal champions rather than mandating top-down usage, and being strategic about where to apply AI in trust-based businesses. Romano emphasizes starting with controlled, limited AI implementations rather than open-ended chatbots, and the critical role of culture change in successful AI workflow adoption.