Software engineering is fundamentally communicative i.e., requirements are negotiated, refined, and validated through natural language. However, many collaborative processes assume neurotypical communication norms, often overlooking the experiences of neurodivergent software engineers such as autistic or ADHD professionals. This position paper argues for neurodiversity-aware conversational requirement engineering, where large language model based agents serve as inclusive mediators during requirement elicitation and validation. By combining advances in semantic parsing, discourse analysis, and adaptive natural language generation, we envision conversational systems that can bridge communication gaps, reduce cognitive load, and foster equitable participation in software development teams.