Title : Nutriagent: Usability and scope of agentic AI towards a personalized, healthy lifestyle and longevity
Abstract:
Every day, millions make dietary promises—commitments to healthier eating that fade within weeks. The difference between nutritional knowledge and sustained behavioral change remains one of health science's most persistent challenges. While people understand what they should eat, maintaining these habits without continuous professional guidance proves remarkably difficult. Traditional nutritional coaching, though effective, remains inaccessible to most due to cost and availability constraints, leaving individuals to navigate their dietary journeys largely alone.
The advent of Large Language Models (LLMs) has ushered in a transformation that may finally bridge this gap. These systems, commonly referred to as Agentic AI, can converse naturally with users, fundamentally changing how we access information and solve problems in both professional and daily contexts. Unlike previous technologies, these agents accept multimodal inputs—text, images, audio, and video—enabling users to interact more intuitively and efficiently. Notably, a paradigm shift is emerging: people increasingly turn to AI agents as their first point of contact for factual information and internet browsing, replacing traditional search engines with conversational interfaces.
This work examines how Agentic AI systems can serve as personalized nutritional guides, addressing the critical gap in continuous dietary coaching. We explore NutriAgent, a carefully calibrated AI system custom-fit to each user's specific characteristics, including personality traits, dietary preferences, and health goals. Users can photograph their meals, discuss nutritional questions through voice input, or engage in text-based conversations—whichever modality suits their moment and preference. The conversational nature enables real-time coaching scenarios: a user can upload an image of their planned meal and receive contextual guidance directing them toward better choices based on their profile and current nutritional best practices.
As AI agents become the primary interface through which people seek information and make decisions, ensuring these systems advocate proper nutritional guidelines becomes paramount. A scientifically curated nutritional agent, grounded in evidence- based practices and personalized to individual needs, can provide the consistent, accessible coaching that traditional models cannot scale to deliver. This research investigates both the usability and scope of such systems, exploring whether Agentic AI can finally transform dietary intentions into sustained healthy behaviors
