AWS has infused its Contact Center Intelligence (CCI) suite with generative AI (GenAI).
While many will be more familiar with its Amazon Connect CCaaS platform, AWS provides the CCI suite to add AI and machine learning (ML) to third-party contact center platforms.
Amongst others, these third-party solutions include Avaya, Cisco, Genesys, and Talkdesk.
In augmenting these platforms, AWS delivers agent-assist, call analytics, and self-service virtual agent capabilities to end-users.
Now, AWS has laced these solutions with GenAI. The following demo from Dr. Andrew Kane, Principal Solutions Architect at AWS, highlights how.
During the demo, Kane notes the additional benefits GenAI delivers as part of the CCI suite:
- Improved call handling accuracy with faster contact resolution
- Script compliance checks and agent scoring
- Task automation that drives cost reduction
- A greater understanding of complex, natural language queries
- Conversational, accurate responses from trusted sources
- Abstractive call summarization
- Lower agent churn
Below is guidance on how it drives these outcomes after an intro into how AWS has embedded GenAI into its CCI suite.
Here’s How the CCI Suite Works
Call analytics sits at the CCI suite’s core. Only by getting to grips with how this works is it possible to understand how AWS is augmenting contact centers with GenAI.
First, note that AWS splits call analytics into live- and post-call analytics. Both have similar architectures – as evident in the following graphic.
However, one significant difference is that – with live-call analytics – it is possible to apply agent-assist. This works by sending the live transcript over to Amazon Lex line-by-line.
From there, Lex interprets the query and Kendra finds relevant information. It then passes this back to Lex, which surfaces those insights on screen, helping agents solve customer queries faster.
AWS has offered such capabilities for months. But now, with GenAI, AWS is taking this further.
This Is Where GenAI Comes In
AWS’s mission in GenAI is to democratize the technology, so companies of all shapes and sizes can access LLMs and build better experiences. As Kane puts it:
We’re not trying to restrict you to one or the other.
As such, AWS provides three methods to augment the reference architecture above with their preferred GenAI model.
These three methods hinge on three different apps: Amazon SageMaker, AWS Lambda, and Amazon Bedrock (as highlighted below).
First, Amazon SageMaker – or SageMaker JumpStart – allows businesses to leverage any LLMs they build within the SageMaker environment.
Next, Lambda allows the company to connect with any other third-party model system – such as one offered by a different vendor or built in an external environment. Lambda acts as the conduit.
Finally, there is the option to utilize Amazon Bedrock. Now, in public Beta, this is AWS’s private GenAI-managed service, allowing businesses to finetune third-party models to their environments.
Each option helps extract the same benefits from the existing CCI suite. Although, those tailored to a contact center environment may deliver better results – as other vendors, such as Observe.ai – have preached.
Enhancing the Value of the CCI Suite
Before generative AI, the CCI suite combined analytics and agent-assist capabilities to deliver contact summarizations that accelerated post-call processing.
How? By pulling lines out of the transcript to recap the thrust of the conversation.
Yet, with GenAI, AWS can generate abstractive call summaries for customers, adding more color to conversations and enabling improved business insights.
Similarly, while the CCI suite previously identified call sentiment, trends, and categories across interactions, with GenAI, it can now classify intent and enable deeper sentiment analysis. It does this through the enhanced NLU capabilities that LLMs make possible.
Again, this evolution helps to enhance business insights, which can feed downstream systems, informing initiatives such as smarter call routing, intent analysis, and quality management.
Regarding the latter, many contact centers currently leverage the CCI suite to assess agent performance and pinpoint learning opportunities.
Yet, GenAI takes this to an even more granular level. As Kane said:
LLMs can answer abstract queries, such as: how did the call go? This is obviously in the transcript, but LLMs are able to look at a conversation in the round and tell you what the call outcome actually was.
Finally, AWS also utilizes GenAI within its CCI suite to support compliance, evaluating call scripts using a knowledge base for the latest policies. That also aids data protection initiatives.
Eager to learn more about how AWS is leveraging generative AI to improve contact center experiences? If so, read our article: AWS Demonstrates How to Augment Lex With Generative AI
