February 20, 2019



A virtual software assistant for banks

Staying in queues in order to talk to a representant of your bank can turn into a waste of time, so how would it be if you could do frequent operations right from your favourite Messenger Client ?

How does it work

In order to extract the intent of messages received from user we use WitAi. It has the following configuration

intent_operation is the core trait, as it encapsulates all the different operation that a user can do:

  • card_renewal
  • block_card
  • check_transactions
  • spending_statistics
  • account_info
  • credit_options
  • show_promotions
  • show_closest
  • add_new_account

Besides that, there are defined all the keywords that an user could write, in our case there are currencies (RON, USD, EUR), sum of money, date


So, when a message is received the pipeline wold look something like this:

  1. send message to witAI
  2. get response
  3. base on the entities provided by witAI (intent operation and keywords) compute and send a response

Technologies used


  • NodeJs
  • WitAi for extracting the meaning of messages


Project link

The Project was realised in the 2nd year of university (June 2018), in a team of 4 during a 48 hours hackathon. I implemented a few of the operations presented above + the interaction with WitAi.