In current customer inquiry and complaint system, customers are tired of waiting on hold in a call or waiting too long for response from the support team.On other hand current chat bots are not advanced enough to take all the actions on their own. This lead to our solution called Smart Resolve.

What it does

This solution contains an intelligent case creation solution called Neo. It automatically tries to resolve the issue with past experience faced by the organisation. If any peculiar question comes up, instead of providing a default answer it passes to the human agent who can take care of the solution. It automatically identifies the question and tries to provide best response for the customer. Even after agent is assigned with a case, it tries to provide response for the queries related to case. This way it helps agent to work only on solution and give their best work

How we built it

This solution is built completely on Appian as appian's one of the favourite area is Customer care center . It makes use of Appian's integration & security system, decision making power, processing engine last but not the least low code platform to build easy UI. Apart from Appian we have chosen facebook as a social networking platform & used google services to provide extended information.

For further details: How To

Challenges we ran into

To make the response real time, we need facebook to call our webApi. This one is tricky because this established through concept called webhook and webhook requires single service which can perform both GET and POST, as well as this needs to be done by Oauth. Apart from that the Get should return a challenge hash to verify which should not be secured. But Appian provides webapi only which are secure.

For a conventional conversation bot, it responds based on single sentence , it doesn't remember the previous conversation. We took this as a challenge, in this solution we try to have a conversation with the user. We even try to perform actions instead of just answering only queries

Accomplishments that we're proud of

All the answers are directly provided from Appian. Even though it uses NLP system to identify question, the answers are dynamically generated from Appian.

The regular chatbots usually can answer only simple queries. However this solution performs some serious action from validating an user in a system, till performing a regular operation in a domain (eg: cancelling ticket). This requires tracking of previous conversation and automatically decides which stage it needs to ask further & which stage it needs to respond with answer. (note: This is not just a trained NLP bot, it is a solution which performs various action based on query, it records the previous conversation unlike trained bots, it answers based on that )

Performing important actions through automated process can lead to serious problems, however we overcome this through OTP concept. This now allows to do all kinds actions without worrying about automation technical error.

Since this is built in a framework format. It should be ready to add new solutions , adopt new domain and integrate with any social platform or organisation's website. This way we try to reduce the human handling work.

It doesnt stop with just answering queries, It helps user from beginning of the case till the end. It plays integral part of whole application instead of just being an gimmick of external services.

What we learned

As part of 18.1 , we learnt to take advantage of appian's new webcontent component which gives rich UI experience. We learnt to make connect Appian with external system through Oauth2.

What's next for Smart Resolve

We are going to extend the framework to support multiple social network and other sources to be the solitary resolution.

Built With

Share this project: