When shopping in Krogers, considering the size and scale of the mart, it is often difficult and time consuming to locate and compare items. For example if I need to find zero sugar cereals or glutten free pasta, I have to go around and compare the items present in the shop. Life would have been a lot simpler, If there was a virtual assistant which can be accessed through web or messaging platforms without the need of installing any applications.

What it does


  1. Helps users to locate items that he wish to shop for
  2. Shows promotions available currently in the shop
  3. Allow users to spot items based on the recipe she/he wants to cook
  4. Helps in logging a service request in case of a emergency or issue
  5. Reads out loud the responses to users
  6. Launches in a web browser session or a messenger bot session by scanning a qr code or visiting url (no need of installation or authenication)

Web Admin:

  1. Allows admin to add items to the knowledge base
  2. Allows admin to add promotions to the knowledge base
  3. Allows admin to add Recipes to the knowledge base
  4. Run batch jobs to do customer profiling for better targeted marketing
  5. Allow admin to create Personas which define different customer profile

How we built it

As our team was having different expertise we built the web services required for web admin part and chatbots using Python Flask. The web admin module is powered by PHP and bootstrap. Chatbot was built with bot builder platform hellotars

Challenges we ran into

We tried hands on Messenger Bot Builder API for chatbots but failed to implement it considering time. Hence had to remove it at the last minute and use another offering.

Accomplishments that we're proud of

Proper running prototypes with expected basic features. Market relevance and Data science capabilities. Simplicity of the offering

What we learned

Bot Building process Collaboration and team work Quick prototyping incorporating a team's idea

What's next for SBuddy

  1. Adding Data science capabilities like clustering and FP Tree for Frequent Item analysis and customer profiling.
  2. Adding NLP features to the chatbot so that customer can ask anything directly without the need to select buttons.

Built With

Share this project: