kitchen cosmo AI prototype transforms leftovers into personalized recipes

Kitchen Cosmo turns everyday cooking into conversation with AI

Developed at MIT, Kitchen Cosmo is a speculative AI cooking device that challenges conventional paradigms of smart kitchen technology. Rather than automating tasks or optimizing efficiency, Cosmo fosters a co-creative relationship between user and machine, generating personalized recipes based on available ingredients and six analog input parameters, including cooking time, mood, and dietary restrictions. The device uses GPT-4o, a multimodal large language model capable of processing both images and text in real time. A webcam captures the user’s ingredients; dials and switches communicate contextual preferences. A single API call then translates these inputs into a context-specific recipe, which is printed via an embedded thermal printer. Cosmo’s distinctive interface is entirely screenless and tactile, rejecting voice assistants and digital displays in favor of knobs, sliders, and physical ritual.

all images courtesy of Ayah Mahmoud and C Jacob Payne

Cosmo’s Red shell and tactile interface nod to retrofuturism

Formally, Cosmo draws inspiration from the 1969 Honeywell Kitchen Computer, a speculative product that famously promised to bring computing into the domestic sphere. With its red retrofuturistic shell and cylindrical recipe archive, Cosmo both pays homage to and critiques that lineage. But where Honeywell’s machine never functioned, Cosmo works, and works differently: improvisational rather than prescriptive, situated rather than universal. This human-centered approach is particularly evident in the novel ‘Cooking Mood’ dial, which allows the emotional character of a moment, including settings such as ‘Nostalgic,’ ‘Spectacle,’ and ‘Surreal,’ to influence the structure, tone, and logic of the generated recipe. Through this reframing, designers C Jacob Payne and Ayah Mahmoud move AI from a tool of standardization to one of interpretation and play.

recipes generated from mood, time, and available ingredients

Kitchen Cosmo device shifts AI from prescriptive to interpretive

Cosmo is part of a broader exploration of ‘Large Language Objects’ (LLOs), physical artifacts that integrate generative models into tangible, interactive systems. While early user testing revealed limitations in cultural bias and recipe reliability, the framework invites meaningful expansion. Future directions include user-specific training, cookbook-informed datasets, and deeper integration of cultural food knowledge. This project by Ayah Mahmoud and C Jacob Payne was developed in the MIT course ‘Interaction Intelligence,’ taught by Professor Marcelo Coelho. By foregrounding improvisation, materiality, and shared authorship, Kitchen Cosmo offers a critical vision for how artificial intelligence might support domestic life, not as an invisible assistant, but as a visible, embodied collaborator in everyday rituals like cooking.

a webcam reads ingredients; dials shape the recipe

GPT-4o powers real-time, multimodal recipe creation

six analog inputs guide Cosmo’s recipe generation

recipes are printed via embedded thermal printer

developed in the MIT course ‘Interaction Intelligence’

Cosmo invites play, interpretation, and shared authorship in cooking

project info:

name: Kitchen Cosmo
designers: Ayah Mahmoud | @_sincerelyayah, C Jacob Payne | @cjacobpayne

instructors: Marcelo Colheo, XDD Dai, Bill McKenna, Sergio Mutis

designboom has received this project from our DIY submissions feature, where we welcome our readers to submit their own work for publication. see more project submissions from our readers here.

edited by: christina vergopoulou | designboom

The post kitchen cosmo AI prototype transforms leftovers into personalized recipes appeared first on designboom | architecture & design magazine.