CalmKitchen Designer

Validate your commercial kitchen design by pressure testing the dinner traffic or walking the setup in 3D. Find gaps before you build and let AI explore competitive alternatives.

Comment

Inspiration

My partner and I build hotels and also write the software to operate them. We're currently building a new hotel and debating how we should order the kitchen equipment for maximum efficiency. We typically use cutouts of the equipment and just move it around the architectural drawing during the brainstorming session.

One of my favorite scenes from the movie 'The Founder' (about McDonald's) was when they drew out the industrial kitchen on a basketball court to help figure out opportunities for efficiencies ("Speedee Service System").

What if we didn't have to do it manually and could actually get better results by running more advanced simulations in parallel? What if the effort to get the results of such a simulation were minimal, due to AI agents already having the context and details to help carry out the busy work with minimal supervision?

What it does

Design your floor plan, fit out your equipment and then simulate peak traffic. It provides a flat view of the plan, a 3D view and a Split view to make changes in your plan while seeing the effect in realtime.

I also added a 'Walk Kitchen' option to get a first person / third person view of walking around the kitchen. This feature has been remarkably popular with my friends, as it gives so much more context about the space than a floor plan or other hand drawn drawing.

There are options to auto-fix issues (equipment not having enough clearance, etc.) at all of the stages. In the simulation, it is possible for the agent to tweak the parameters for better reflection of the actual dishes used in service. The sim mode also provides a service scorecard and the option to compare the performance of different layouts.

Since this is a tool we're leaving open for restauranteurs/hoteliers, it requires no login and allows users to save their layout with changes locally.

How we built it

We used our actual floor plans to help drive the requirements. Since my partner is not an engineer, she prefers to use ChatGPT's in-built browser with WebMCP to implement the changes / drive the results with minimal manual work.

The base product had revisions built into it. We went from 2D -> 3D -> Simulations. The project was done in a little over 2 days.

Challenges we ran into

Spatial conflicts, in term of clearance for opening freezers, etc. and adding validation in both the 2D and 3D spaces. WebMCP within ChatGPT was also interesting in the sense that it did not check for availability and had to be explicitly informed about it.

Accomplishments that we're proud of

'Walk Kitchen' option. It's basically the same as playing Doom with a spatula. Building a restaurant is hard - let us have a bit of fun :-)

What we learned

Small changes go a long way, when we impose additional modifiers such as staff calling in sick, etc.

What's next for CalmKitchen Designer

Our core product is HotelOS.ai, with CalmKitchen running as the operating layer for restaurants that are on-prem. We'll continue improving CalmKitchen Designer so that hoteliers and restauranteurs can hopefully save a few of their nine lives before they step into the arena.

Built With

Share this project:

Updates

Submission history