Chefmate: AI-Driven Kitchen Operations

About the Project

Chefmate is an AI-driven solution for professional kitchens—helping with weekly menu planning, grocery ordering, and cost monitoring. I was brought in to convert a loose requirements document into actual userflows and wireframes, working directly with the customer to understand their needs and shape the product for an upcoming pilot.

Understanding the Customer

Through workshops and video meetings, I worked to understand how professional kitchens actually operate. The conversation revealed what mattered most:

For Head Chefs (kjøkkensjef):

• Assist and increase their creativity in menu planning
• Inspect individual meal recipes for each day
• Alter ingredients for specific meals, entire days, or the full week
• Generate multiple menu versions to choose from

For Chefs:

• Receive generated menus ready for selection
• Have that selection automatically converted to a purchase order
• Use that order in the existing ordering portal to buy groceries and supplies

For Area Managers:

• Oversee key statistics from each kitchen
• Access regional data to identify patterns and opportunities

For the System:

• Simple user management to assign employees, kitchen managers, and regional managers to kitchens
• History of weekly menus for reference and analysis

From Requirements to Design

The initial brief was loose. My job was to translate that into clear userflows and wireframes that could guide development. This meant asking hard questions about what "generate multiple versions" actually meant, how chefs would interact with AI suggestions, and where human judgment needed to override automation.

The workshops were essential—they revealed assumptions in the brief that didn't match kitchen reality, and clarified where the AI assistance added real value versus where it would just create friction.

Into the Pilot

Once the flows and MVP were locked, I moved into design finalization and then into a hybrid role: part product manager, part prompt engineer. This meant:

• Tweaking AI settings and dialog based on user feedback from testing
• Collaborating with the developer on technical challenges
• Working with the customer on iteration and refinement
• Problem-solving around constraints imposed by the hosting solution

The work required moving fluidly between design, product thinking, and technical collaboration—understanding enough about the constraints to find workable solutions rather than just asking for the impossible.