Background: Food security and maternal nutrition are national priorities in stunting prevention programs, particularly in Bima District, West Nusa Tenggara, where the prevalence of stunting reaches 29.5%. Although Artificial Intelligence (AI) offers great potential for supporting nutritional interventions, the main challenge lies in its limited adaptation to cultural diversity that influences dietary patterns. Objective: This study aimed to develop and evaluate the effectiveness of the culturally adapted SandoAI platform in improving nutritional literacy and food security among pregnant women. Methods: The research applied a Research and Development (R&D) design using Borg & Gall’s model, involving 200 pregnant women, 40 health workers, and 15 stakeholders. The instruments included a 20-item nutrition literacy test, a 15-item food security questionnaire, interview guidelines, and observation sheets. Data were analyzed using descriptive statistics, paired t-tests, N-Gain, and qualitative descriptive analysis. Results: Expert validation produced an average score of 4.2 (very feasible), while the limited trial resulted in an average score of 4.25 (very good). Effectiveness testing showed significant improvements in nutritional literacy and food security with N-Gain values ranging from 0.62 to 0.65 (moderate category) and p < 0.05. Conclusion: SandoAI is considered feasible, practical, and effective as a culturally based nutrition education platform. It holds strong potential as a strategic innovation for stunting prevention by integrating AI and local wisdom in an adaptive and inclusive manner, aligning with national digital health transformation policies.