How Contextual Answers Is Transforming Customer Support

AI21 Labs's Contextual Answers is a question-answering-based Generative AI that answers based on a company's knowledge base and uploaded documents. Support agents can ask questions in natural language to quickly find the information they need and provide customers with more accurate and personalized responses.

This approach is used in two ways.

• Internal support efficiency: Reduces the time agents spend on search and research, improving response speed and first-call resolution.
• External automated response: Attached to a website's chatbot or search bar to immediately handle common inquiries, escalating only complex issues to humans.

The article addresses the cost, processing speed, and self-service limitations of customer support, and explains that Contextual Answers can alleviate these issues. In particular, it emphasizes that answers are designed to be based solely on uploaded materials, reducing AI hallucinations, and that providing the source of each answer together with the response increases trust.

The implementation process is organized into four stages. First, determine whether the goal is internal or external use, and decide on language support and LLM selection criteria. Next, collect and label documents so that different answers are generated depending on customer tier or product, then connect the model and documents and deploy.

Continuous evaluation is important during operation. The article cites a digital bank's case, describing how they built a system to score answers on a 1-5 scale and managed quality. As a result, they achieved improved customer satisfaction while being able to operate with a smaller support team, also gaining cost-saving benefits.