The copilot is being designed to use computer vision and AI-generated prompting to help users prepare healthy meals.<\/li>\n<\/ul>\nLOS ANGELES \u2014 Researchers at Stanford Medicine are developing an AI cooking copilot to help people make healthy and flavorful meals. <\/p>\n
The project is led by Minal Moharir, MD,<\/b> a clinical assistant professor of medicine, primary care and population health at Stanford Medicine. <\/p>\n
Moharir told Healio it is often challenging for patients to change their cooking habits for healthier, plant-forward meals. One reason is the lack of guidance on incorporating cultural preferences. This prompted Moharir to launch the Nourish project at Stanford Medicine, which aims to develop culturally tailored resources for preventing and managing diabetes.<\/p>\n
\u201cWe mainly want to create these resources so that patients can continue to eat the foods they love,\u201d she said. \u201cOur main goal is to not really tell our patients what not to eat but give them guidance in what they can eat.\u201d<\/p>\n
One initiative under the Nourish project is a teaching kitchen program that combines expertise from physicians, registered dietitians and chefs to educate participants about healthy cooking techniques for foods that they will enjoy.<\/p>\n
The teaching kitchen program emphasizes the importance of a recipe\u2019s sequence, Moharir said.<\/p>\n
For example, \u201cif the oil doesn\u2019t shimmer, then the mustard seeds are not going to pop, then the ginger is not going to sizzle, and maybe the dish is not going to be delicious,\u201d she explained.<\/p>\n
Moharir and colleagues held two teaching kitchen sessions for 16 Stanford employees. The sessions were well-received, with most participants rating it as \u201cexcellent\u201d and \u201cextremely useful,\u201d and many requesting additional programming, according to a poster presented at Cedars-Sinai\u2019s Virtual Medicine Conference.<\/p>\n
The study was confined to 16 people due to limited staff and resources, but 27 people wanted to participate in the program, according to Stefan Thottunkal, BHLTH, MS,<\/b> a researcher for the Nourish project. The program\u2019s popularity led to the \u201cvision behind the AI copilot,\u201d he told Healio.<\/p>\n
Researchers at Stanford Medicine explain why and how they are developing an AI cooking copilot.<\/figcaption><\/p>\n\u201cWe realized that if we could take some of these insights that we\u2019re gaining and feed them into a [large language model], we could provide personalized guidance and use computer vision to provide feedback as people cook,\u201d he said.<\/p>\n
Moharir, Thottunkal and colleagues are designing the AI cooking copilot to deliver real-time, step-by-step support, from recognizing the right pan temperature to guiding ingredient timing, sequencing and flavor-building techniques. <\/p>\n
\u201cIt can say at each step that you complete what the next step should be,\u201d Thottunkal said. \u201cCooking is a sequence-dependent biochemical process. The way fat is heated changes extraction of hydrophobic flavor molecules from spices and aromatics. The way ingredients are browned influences Maillard-derived flavor development. The way starch-rich ingredients are cooked alters gelatinization, digestibility and potentially postprandial glycemic behavior. So, when we talk about step-by-step cooking support, we are really talking about guiding the chemistry that determines taste, texture and health relevance. That is where an AI copilot can be powerful, because it can help translate food chemistry into practical, culturally relevant guidance in real time, so you don\u2019t need to know the chemistry to enjoy the best possible outcome, just follow the steps!\u201d<\/p>\n
The AI copilot is now moving through active development, Thottunkal said.<\/p>\n
\u201cWe\u2019re analyzing the teaching kitchen findings and figuring out how we can use prompt engineering, retrieval-augmented generation and efficient data pipelines to turn that into really practical guidance for someone with limited cooking experience,\u201d he said. \u201cFrom there, we\u2019ll move to demos and testing with real people in the lab.\u201d<\/p>\n
For more information:<\/h2>\n Minal Moharir, MD<\/b>,<\/b> <\/b>and Stefan Thottunkal, BHLTH, MS,<\/b> can be reached at primarycare@healio.com.<\/p>\n\n
\n
Sources\/Disclosures<\/h3>\n Source: <\/h2>\n Healio Interviews<\/p>\n
Reference:<\/h2>\n\nMoharir M, et al. Participant motivations and learning from a culturally tailored teaching kitchen: Informing an AI cooking copilot. Virtual Medicine Conference; March 25-26, 2026; Los Angeles.<\/li>\n<\/ul>\n\n
\n Disclosures: <\/strong>\n
\n
<\/p>\n
Ask a clinical question<\/strong> and tap into Healio AI’s knowledge<\/strong> base.<\/p>\n\nPubMed, enrolling\/recruiting trials, guidelines<\/li>\n Clinical Guidance, Healio CME, FDA news<\/li>\n Healio’s exclusive daily news coverage of clinical data<\/li>\n<\/ul>\nLearn more<\/button>\n<\/div>\n<\/div>\n