Artificial Intelligence (AI) continues to permeate different sectors of human societies across the world. However, there is growing concern that current patterns of AI technology adoption are likely to exacerbate existing inequalities, with undesirable consequences for underserved communities. AI is also seen as a potential solution for addressing these very inequalities. However, little is known about how innovative AI interventions may be introduced in low-income settings to address societal challenges. In this work, we present our efforts towards designing a community AI intervention for scaling up adolescent access to sexual and reproductive health and rights (SRHR) education. We employed a participatory approach to design a prototype of our AI intervention based on synthetic data, which aided gathering of requirements, experimenting with user experience design (UXD), and stakeholder engagement with content. We found that participatory AI design approaches that allow direct engagement between designers and stakeholders were more beneficial in identifying community needs. We also found that our prototype, called TeenApp, was effective for engaging with users on UXD and AI-generated SRHR education content. However, challenges abound when implementing community-based AI projects in low-income settings. These include lack of AI projects to benchmark from, limited resources, misplaced community expectations, and lack of AI dataset.