A leading contact center application provider based in Indonesia offers innovative solutions that enable seamless communication between customers and businesses. With a focus on leveraging cutting-edge technologies, this provider empowers companies to deliver exceptional customer experiences while optimizing operational efficiency.

Business Challenge

The contact center application enables customers to connect with companies via voice calls or chat without using call credits, leveraging only an internet connection. The application also offers features like location sharing, product information viewing, and direct contact center connectivity.

At the contact center, multiple agents are available to assist clients. However, a significant challenge arises when agents struggle to provide responses that align with the customer’s expected standards. While agents take notes during kickoff meetings to understand the desired manner of addressing certain inquiries, they occasionally face difficulties in locating the appropriate information and determining the optimal way to convey the answer to the client’s customers.

Technology Challenge

The contact center agents often encounter challenges when responding to inquiries that involve specific conditions or scenarios. In certain instances, the agents may lack the preferred knowledge or expertise to provide accurate and appropriate answers. To mitigate the risk of delivering incorrect information, a platform was required that could offer recommendations from a GenAI virtual assistant specifically designed for contact center assistance.

Solution

The solution was to create a virtual assistant powered by AWS Bedrock for contact centers, enabling the team to receive insights and summarizations from the GenAI Contact Center Virtual Assistant based on the knowledge base or via chat interface.

The solution employed a few-shot prompting by providing the Generative AI models, Claude v3 Haiku and Sonnet, with a curated set of examples from the customer support templates, guidelines, and processes. This approach allowed the models to swiftly comprehend the context and generate relevant and tailored responses. Furthermore, an advanced Retrieval Augmented Generation (RAG) implementation was seamlessly integrated, empowering the models to dynamically retrieve and assimilate pertinent information from the knowledge base during the RAG process, ensuring comprehensive and up-to-date responses.

To enhance the retrieval and summarization process, Cohere’s Multilingual Embedding Model v3 was employed to generate vector representations of the knowledge base, encompassing both Bahasa Indonesia and English content. These vector representations were then securely stored in an Amazon Aurora PostgreSQL database, leveraging the pgvector plugin. This innovative approach facilitated efficient and accurate retrieval of relevant information during the RAG process, ensuring that the GenAI Contact Center Virtual Assistant could provide comprehensive and language-agnostic responses to customer inquiries.

On Responsible AI

Following a Responsible AI Policy, the use case for the contact center operations was designed, adhering to the principles of Transparency, Reliability & Safety, and Human-Centricity. In line with Transparency, the use case incorporates clear human oversight throughout the process, with agents maintaining control over AI-assisted interactions. The Quality Check step ensures Reliability & Safety by rigorously testing and validating the chatbot’s responses for reliable and safe operation.

Furthermore, the use case exemplifies Human-Centricity by adopting a human-centered design. The AI chatbot augments and supports human agents rather than replacing them, considering the broader societal impact of AI on employment and customer service. This approach strikes a balance between technological advancement and human involvement.

Customer Outcomes and Benefits

The GenAI Contact Center Virtual Assistant solution was adopted by the entire team, consisting of multiple agents and supervisors. This solution streamlines the customer support process for numerous inquiries annually across operations in Indonesia and helps provide more efficient and accurate responses.

During the initial testing phase of the Contact Center Virtual Assistant, feedback from 5 out of 6 customer service agents reported that they found the Assistant helpful. The virtual assistant has facilitated the professional development of contact agents, serving as a valuable brainstorming tool and enabling engaging training activities that foster skill enhancement.

The reporting capabilities have been elevated, empowering clients with access to the most frequently asked questions and recommended responses, thereby enabling them to deliver more informed and tailored support to their customers.

Their IT Manager stated, “The top-performing GenAI Virtual Assistant has enabled our Contact Service Team to provide answers in an accurate and preferred manner, thereby enhancing customer engagement and benefiting our business operations. Previously, our Contact Center team lacked specific guidelines on maintaining a consistent tone or responding to particular situations. However, with the GenAI’s implementation, it can now offer real-time suggestions tailored to each customer’s inquiries, ensuring a more standardized and appropriate approach to addressing their questions.”