For companies/enterprises to work efficiently and maintain a healthy relationship with their customers it is of utmost importance to solve the technical difficulties of customer as fast as possible. However, for companies receiving a lot of these complaints it becomes extremely difficult to organize these requests efficiently and leads to the wastage of a lot of time. To solve this problem I created this solution.
What it does
It uses a QB app and pipelines to store all the data for a customer issue ( via slack and email) and then uses ML to detect language, extract keywords from the issue being generated and determine if the query is something urgent or not.
How I built it
The solution involves slack pipeline where if a user mentions the name of the bot only then the issue is created in the QB app. A confirmation message is immediately sent back to the user via slackbot and then the ML pipeline is called where the language is detected, keywords are extracted and the issue is deemed to be urgent or not urgent. In the email pipeline, as soon as an email is sent to the defined address, we receive an email of confirmation and the mail is sent to mailparser.io where it is parsed accordingly. As soon as it is parsed it triggers the pipeline which consists the mailparser channel and the issue is created in the QB app, the ML pipeline is called and it inserts all the issue with all the extra data into the QB app.
Challenges I ran into
Understanding how quickbase works (First time user) and deciding on a topic.
Accomplishments that I'm proud of
Finishing this entire project solo with some time to spare
What I learned
Quickbase pipelines & apps ( and the struggles of the customer support industry)
What's next for Support Ticket Manager
Once quickbase pipelines allow for the extraction of images from slack and email directly, users will be able use attachments for their queries when creating support tickets .
The keywords can be used efficiently with a deep learning model( personalized ) to allow for the auto tagging of customer issues. (For e.g. tags such as "price" is more frequently used in billing problems rather than bug problems)