Messenger bot giving you food suggestions
View Process Deck
Type and Timeline
Facebook Hackathon
Nov 1-2 2019
My Role
Product Design
User Research
Will Kim
Derick James
Renaise Kim
Shubham Gupta

During the 10 hour Chicago Developer Community Hackathon from November 1-2, 150 attendees formed to build innovative products for the future using Spark AR, Messenger, and React/React Native. Teams were measured on quality, implementation, development, and potential impact of the product.

Among the three given categories of Social Good, Productivity and Utility, and Entertainment, we chose to focus on the Productivity and Utility category.
For the backend, we programmed with Ruby to talk to our Messenger API. We built the product page with React and then deployed the final version on Github. We used Sketch for the design assets and slide deck.
What do I eat today?
Hunger is an emotional trigger and can heavily influence people’s decision making. This can lead to irrationally scavanging whatever is easiest to consume to alleviatethe grumbling monster that is your stomach.
Time is the most valuable currency. On average, the American spends over 240 hours per year on their meal deciding journey. If the average American eats 3 meals a day, the time surrounding the event can be streamlined through a more efficient framework.
How do we streamline a personalized decision tool that emulates a friendly dialogue and advances efficiency throughout the whole experience from the decision state to filling your belly with localized options?
Why use Munchbot?
In other words, how is this different from Grubhub or Yelp?
Munchbot extends from a personalized suggestion framework, by removing the heavy consumerist interface of traditional localized search and discovery apps like Yelp or the discovery feature on Google Maps, Munchbox provides the most complex asset which is simplifying the discovery experience through a natural and delightful dialogue.
Introducing our avatar, Munchie. The design of the avatar reflects our product values; human centered, delightful, and amicable.
The devices for Munchbot will be available on desktop and mobile. Therefore, the key features should result in a seamless and personalized onboarding experience by prioritizing distance, budget, and food compatibility.
Online and Mobile Device
Key features of the bot dialogue
1) Budget
2) Distance
3) Food compatibility
    a. Favorable cuisine options
    b. Dietary restrictions
User Privacy
Personalization includes the storing of data, but Munchbot respects how that is utilized and does not contribute to advertisement that is seen on Facebook and elsewhere.
Utilize Munchbot's dialogue to authenticate information that includes
a. Food preferences
b. Dietary restrictions
c. Realtime location via "current location"
We do not store sensitive data and data is stored such as name, phone number, etc.
Personalization includes the storing of data, but Munchbot respects how that is beingutilized and does not contribute to advertisement that is seen on Facebook and elsewhere.

Utilize Munchbot’s dialogue to authenticate information that includes:
1) Food preferences
2) Dietary restrictions
3) Realtime location via "current location"

We do not store sensitive data and data is stored such as name, phone number, etc.
User flow
Product Website
The product website was built with React and utilizes a built in Messenger bot chat module that is connected to the user's personal Facebook Messenger account. This website can be visited here.
The user can look up the search query "Munchbot" in the Messenger search tab or can automatically initiate the chat through the Facebook page. Munchbot offers a warm primer to collect appropriate personalized information without sounding too robotic. The inital question activates the type of cuisine filter.
With the product main features in mind, Munchbot asks for the price point to further filter the local options. Similar to Google Map's price point signifier, the options $, $$, $$$, and $$$$ are offered. A numerical value can also be entered and the bot will correspond according to the digit value. To provide quick options, Munchbot gives the user 3 options
Advanced Personalization
After the user selects their desired meal, the bot will generate the options nearby along with the distance of the commute the user desires. In this user flow, the user selects less than three blocks. The dialogue box "Want more options?" will pop up after the three initial local eatery options. When the user selects this, they can select additional options including Google Maps, which will lead them to the selected option. Munchbot will then ask for feedback promptly after the commute time. The user can then log this information to Munchbot which can allow for personalized affordances, facilitating the recommendation process.
We are very proud to commence Munchbox winning the Productivity and Utility category. One of the members, Will Kim, an incredibly talented developer, has went beyond the hackathon and is currently building an API for Munchbot to make the product stronger. More details available on
What we learned
1) Researching the scalability of the product, user privacy, data storage, and adapting to the Messenger API.
2) Programing without getting tangled in natural language proccessing within the short timeframe.
3) Designing with the customer requirement analysis framework in our case and coming up with an interface which enables minimum text input.


1) Integration of recommendation engine built using python with Ruby messenger app.
2) We initially started implementing the backend with node.js to which we were not able to fully implement the idea as we faced some technical challenges.

Next Steps
1) Add poling functionality for a group to vote for a place to go.
2) Venmo functionality to split costs.
3) Collect training data for better personalization.
4) Recommendation engine based on NLP on reviews.
5) Personalize experience (Parking preference, Payment,preference, Pet preference, Drive thru, Discount deals, Music - Karoke, Live, By appointment/no appointment)
Thank you for viewing the Facebook Hackathon project! For any feedback please email me at
View More Work

(c) 2020 Renaise Kim