Enhancing Total Experience in Contact Centers with TCS solution Powered by Generative AI on AWS

By Kathirvelan G, Product Lead, AI.Cloud — TCS
      Muthukumaran Krishnan, Solutions Architect, AI.Cloud– TCS
      Dhiren Mehta, WW Leader for Partner Solutions Architecture, AWS
      Nitin Chahar, Principal Solutions Architect, AWS

TCS

Contact centers serve a diverse customer base spanning different strata. While GenX customers prefer interacting with contact center agents, either in person, over live chat or phone, millennials utilize a combination of internet, email, text, and voice channels to address their queries and get their issues resolved. GenZ customers have fully embraced digital channels, including a multitude of social media platforms and voice assistants like Alexa. This cohort also desires the ability to seamlessly continue a conversation across different channels, as they often engage with contact center agents while multitasking to accomplish other work. Leveraging the power of generative AI, Contact Centers can hyper-personalize to evolving customer preferences, while dynamically providing agents relevant context to increase first-contact resolution and enhance customer satisfaction.

The primary customer interaction point in contact centers is through self-service, facilitated by conversational chatbots that are accessible across various digital channels, such as customer portal, and mobile applications. Contact center agents continue to represent the second tier of customer support, after chatbots, and are responsible for providing personalized service for issues that require human intervention. It is essential to equip agents with tools that enable them to provide accurate information in real-time during customer interactions. Further, providing agents with tools to expedite their post-call activities enhance their efficiency. Additionally, contact center managers and quality analysts require performance measurement tools to optimize agent training, staffing, and quality — reducing costs while improving customer service.

Contact centers should be accessible across multiple channels to accommodate diverse customer preferences and to provide customers a uniform and seamless experience across the channels.

In this blog post, you will learn about the challenges and expectations facing contact centers, along with Tata Consulting Services (TCS) innovative solution that leverages Generative AI on AWS to transform these operations.

Tata Consultancy Services (TCS) is an AWS Premier Tier Services Partner and Managed Services Provider (MSP) with AWS Generative AI competency. TCS has been partnering with many of the world’s largest businesses in their transformation journeys for the last 50 years.

Challenges

Contemporary contact centers encounter the following challenges:

• Due to high overheads conducting quality assurance (QA) manually, it is selectively applied to a small sample of customer calls, which fails to deliver a holistic understanding of the contact center’s overall operational performance and effectiveness.
• Agents lack contextual information to answer customer queries and tools for training and performance feedback.
• Customer Insights like customer satisfaction level with specific function, channel of choice, frequent reason for contact and proactive communication strategy are not readily accessible to business stakeholders.
• Agents dedicate a significant portion of their time to summarizing call recordings for note-taking purposes.
• Traditional chatbots often struggle to understand the semantic meaning of customer queries and lack robust support for multiple languages and translation capabilities.

Expectations from contact centers

Following are some key expectations from contact centers to provide seamless experience to the customers:

• Deliver consistent, high-quality service to customers regardless of the onboarding channel utilized.
• Facilitate seamless omni-channel interactions, enabling customers to engage through their preferred channels.
• Transition to real-time call analysis to enhance customer experience and drive operational efficiencies.
• Implement a value-based selling approach to better understand customer needs and provide tailored solutions through cross-sell/upsell.
• Chatbots ability to leverage semantic search and multilingual translation to comprehend customer intent and serve diverse global users effectively.

Why Generative AI?

The effect of Generative AI will quickly deliver new benefits and astounding progress, which are as follows:

• Generative AI focuses on generating Output in natural language in conversational tone better suited for customer interaction
• Generative AI can help personalization of interactions with customers.
• Generative AI powered virtual agents can demonstrate empathy and address customer queries by enabling them to communicate in the language of the customer.
• Generative AI learns from each customer interaction and provides increasingly accurate and relevant information through refinement process.
• Generative AI enables semantic search, offers multilingual support with capability for live translation.

TCS Reference Architecture for Generative AI on AWS-Powered Contact Center solution

The reference architecture harnesses the power of Generative AI to augment the productivity of contact center agents by providing them with real-time, contextual responses to customer queries, and guided dialogue, and automated call summaries. This reduces Average Call Handling Time (AHT) and optimizes overall efficiency.

From customer standpoint, this solution offers seamless self-service capabilities through advanced chatbots with natural language processing, addressing their queries and enhancing self-service portals with contextual responses from enterprise knowledge, streamlining interactions and improving overall customer experience.

For employees managing the contact center, it offers automated call quality assurance, effective issues resolution, unsolved issues escalation mechanism, and delivers insights to improve customer satisfaction to improve the Customer Satisfaction (CSAT) score.

The composable nature of the solution ensures that customers can selectively implement the features that align with their unique requirements, providing a flexible and adaptable solution.

 Figure 1: Solution Tenets

Solution Architecture

Let’s go through the solution components in detail, including AWS services leveraged for each component. The key AWS service leveraged in this solution is Amazon Bedrock, a fully managed service that offers a choice of high-performing Foundation Models (FMs)/Large Language Models(LLMs) from leading AI companies.

   Figure 2: TCS Reference Architecture for Generative AI-Powered Contact Center

Self-Service Portals

Customers are presented with a Question & Answer(Q&A) interface, which they can use for asking questions in natural language and in turn getting a response via Amazon Bedrock API (leveraging Anthropic Claude V3 Sonnet LLM). This response has been contextualized using data from Amazon OpenSearch vector database to understand and generate human-like text based on the input they receive.

Language Translation

Anthropic’s Claude V3 Sonnet LLM enables seamless two-way translation and deep comprehension of customer intent and context, facilitating effective cross-lingual communication between agents and customers

GenAI Chatbots

Customers can access the Generative AI-powered chatbot, which leverages Anthropic Claude V3 Sonnet LLM via Amazon Bedrock API, through various channels like customer portal and mobile apps. This ensures a consistent self-service experience for customers, regardless of the communication channel they use to interact with the chatbot.

Customer Survey and Feedback

Customer feedback surveys are sent to customers after a contact is completed from any channel through the service. The survey can be administered in a language of customer’s choice. The responses from the survey are analyzed by the Anthropic Claude V3 Sonnet LLM, and the insights generated are stored in the Amazon DynamoDB, which are then available to managers through the survey dashboards.

Real-time Sentiment Analysis & Emotion detection

The agent’s capability to get real-time sentiment analysis is achieved by passing the transcripts to the Anthropic Claude V3 Sonnet LLM. The prompt engineering, is used to understand the customer’s emotion and provide advice, which is then relayed back to the agent using AWS Lambda functions for proactive customer engagement.

Smart Search & Knowledge Management

The chatbot provides agents with a smart search capability powered by Amazon Bedrock Knowledge Bases, a fully managed Retrieval-augmented generation (RAG) implementation. This establishes a robust knowledge management by processing knowledge articles, making them easily available for searching using OpenSearch index.

Call Quality Assurance

The call details are streamed using Amazon Kinesis and stored in Amazon S3. Subsequently, by leveraging Amazon SQS, these scripts/recordings are input into the Anthropic Claude V3 Sonnet LLM (via Amazon Bedrock API) for analysis and summarization. The helps in identifying the issues and resolutions to determine if the calls were handled correctly by the agents as defined in the playbook. Notifications can be sent using Amazon SNS to managers and key stakeholders after the call quality analysis has been performed.

Call Summary

After each call concludes, the system generates a call summary by processing the transcript through Anthropic Claude V3 Sonnet LLM. This summary empowers the manager to promptly assess whether the call was resolved appropriately by the agent.

Customer Insights

Anthropic Claude V3 Sonnet LLM is used for topic modeling to identify and categorize the main topics or themes of customer queries. This allows contact centers to understand the most common reasons customers reach out and address trending issues.

Responsible AI

Customer conversations while they are interacting with the virtual agent are kept on topic by implementing the topic filter using Amazon Bedrock Guardrails. With content filters and word filters, usage of inappropriate language and harmful content is prevented. With contextual grounding, hallucinations in model responses in model responses are filtered. With PII data redaction, sensitive information is safeguarded.

Solution Benefits

The composable nature of this solution empowers customers to choose and customize features according to their specific requirements. It offers the following benefits to customers:

• Enhances customer satisfaction by 25% and increases customer retention.
• Improves agent training effectiveness by 30%.
• Self-service bot handles incoming calls and resolves issues, enabling agents to focus on personalized service and reducing operational expenses for contact centers.
• Enhances agent productivity by up to 35% through streamlined processes, effectively reducing average call handling time.

Conclusion

Generative AI is revolutionizing contact centers, enhancing customer interactions and operational efficiency. TCS’s solution leverages this technology to provide a comprehensive, multi-device experience for customers, agents, and employees. It streamlines operations, improves agent productivity, and delivers superior customer service. This AI-driven approach enables contact centers to deliver superior customer service while driving operational excellence.

If you’re interested in transforming your contact center with Generative AI or learning more about this solution, please contact TCS to schedule a demo.

.

TCS — AWS Partner Spotlight

TCS is an AWS Premier Tier Services Partner and MSP that has been partnering with many of the world’s largest businesses in their transformation journeys for the last 50 years.

Contact TCS | Partner Overview | AWS Marketplace