Our client is a leading financial services company who provide advice and guidance to both consumers and businesses. They had a requirement to replace their existing telephony system with a more future proofed solution which allowed them to better understand the nature of the call, to provide greater self-serve capabilities and to streamline call routing via matching spoken phrases to specific call intents.

The requirement to Alscient consisted of the following:

• To provide a virtual assistant which allowed the caller to specify the nature of their enquiry. Utterances were to be matched to five distinct intents with the call being routed either to an internal team or to an external company based on the nature of the enquiry.
• The solution needed to pop the call, Case and Account within Salesforce.
• The call needed to be routed to a specific team based on the company classification type associated with the caller.
• The solution needed to support the ability to transfer the call to another team or external third party.
• The virtual assistant also needed to offer out a route from the IVR if the caller was struggling to match to a specific intent or if the caller wanted to speak to an agent.
• The solution also needed to provide the ability to record the call and also screen record the session for training and quality assurance purposes.
• Advanced analysis of the call was also to be conducted to ascertain the call sentiment, to assess for call quality outcomes and also to provide a purpose for the call.
• The solution needed to provide real-time dashboarding capabilities to show a view of inflight calls being received.

We addressed these requirements in a phased approach. Initially captured the full set of requirements for the solution and provided a high-level design which described how the solution would work and the supporting technical architecture. Following sign-off of the design, the solution was rapidly prototyped with customer feedback being received via a series of show and tells. Intelligent routing made use of data dips into Salesforce based on the callers number and also based on DTMF input.

The integrated Salesforce solution was based on Service Cloud Voice which also made use of Amazon Connect, Amazon Lex, AWS Lambda and supporting services. A Lex bot and a series of Lambdas were used to provide the virtual assistant. Hundreds of utterances were matched to each intent and a confidence score was used to assist with intent matching. If the accuracy passed a pre-defined threshold the call was linked to a specific intent and the spoken phrase along with the intent matching score were stored against the created Case. Some of the matched intents resulted in calls being routed to internal teams whilst others resulted in the call being directed to external third parties which reduced the number of calls into the contact centre.

A close all lines override was also built into the solution making use of Dynamo DB and Amazon Pinpoint. Joulica was used for real-time reporting to provide a view of inflight calls. S3 was also used to store the call and screen recordings and Polly was used to provide text-to-speech prompts within the call flow.

Following completion of build the solution was system tested and then migrated across to a UAT environment where our client performed their own testing of the solution. After completion of UAT the solution was then promoted to the live environment.

The solution has been live since December 2023 and allows customers to self-serve with many common enquiry types without needing for the request to be serviced by an advisor. The solution has also provided much deeper insight into the nature of the calls with Contact Lens being used to assess call sentiment, provide a gist for the call and also to assess for different call quality outcomes. Our client indicated that this was one of the top five projects they undertook last year and were very happy that the solution was delivered to their aggressive timeline.

Our client is now looking to further extend the solution, making further use of supporting AWS services, including the use of private hosted generative AI models which work alongside Amazon Connect and Amazon Lex.