Despite their pivotal role in integrating AI into education, instructors’ adoption of AI-powered tools remains inconsistent, with limited research on designing AI tools for broader adoption. This study employs a human-centered design and qualitative approach to investigate the design of interactive pedagogical agents that provide suggestions on instructors’ questions. First, a formative study examined current pedagogical support strategies through interviews with five pedagogy experts. Second, a participatory design session engaged ten pedagogy experts in reviewing a storyboard illustrating chatbot features tailored to instructors with varying AI literacy and attitudes. Experts also evaluated the quality of LLM-generated suggestions based on instructors’ common teaching challenges. Findings highlight the need for chatbot interactions that foster trust, particularly for AI-conservative instructors, by providing social transparency in peer usage and greater flexibility in how much or how little instructors engage with the system. Additionally, we propose design recommendations for enhancing AI-generated teaching suggestions, such as assessing prior teaching experiences. This work underscores the urgency of supporting AI-conservative instructors as AI literacy and attitudes intertwine, posing risks of pedagogical divides and diminished student learning opportunities if left unaddressed.