Large language models (LLMs) have exploded in popularity since the advent of ChatGPT. Digital First Initiatives (DFIs) have been touted as a likely home for LLM based products due to their prospective versatility at handling patient engagement, and reducing related pressures on patients and clinicians alike. This study employed a mixed methods approach to capture the perspectives of stakeholders on each side of the DFI divide. Surveys targeted medic, and non-medic student samples for informative descriptive statistics. Focus groups informed the qualitative analysis, recruiting 6 medics, and 6 non-medics respectively, supporting implementation of Thematic Analyses and Line of Argument Syntheses. The combined findings pointed to the deeply embedded frustrations with traditional pathways to care, and a desire for more efficient digital means. Dualistic preferences over the interactive qualities of LLMs highlighted a strong need for granular personalization efforts, with general preference for a balance between LLMpathy and system efficiency. The outcome of the Line of Argument Synthesis generated five conceptualizations of stakeholder concerns, potential paths to mitigate concerns, and the mediating impact this would have on LLM uptake in DFI Pathways. The study supports efforts to explore successful integration of LLMs in Digital First Initiatives. Careful attention should be given to ensuring stakeholder perspectives are taken into account at all stages of design and implementation.