Contact-center-as-a-service provider Five9 Inc. today said it has expanded the Genius AI suite it announced in August with intelligent virtual agents that incorporate generative artificial intelligence and natural language processing to simulate the conversational abilities of a human.

Businesses can fine-tune Five9 AI Agents to balance flexibility and control by adjusting the amount of generative AI and traditional pre-defined AI used in the same conversation.

Citing its recent research that found that 56% of U.S. and U.K. customers are frustrated by self-service chatbots, the company said its new agents allow responses to be scripted when needed or flexible depending on the customer’s needs.

Many chatbots rely on scripted responses and limited knowledge bases, making them unable to handle complex, fuzzy or dynamic requests. Five9 said its new AI Agents’ descriptive design significantly reduces the complexity and cost of enabling customer self-service. Descriptive design helps establish a clear, factual understanding of a subject by systematically describing characteristics, behaviors and patterns.

Years in the making

Five9 has been working with AI for years. In 2018, it launched Genius, an AI-powered contact center tool that uses data from emails, chats and web searches to predict why someone contacts customer support and match them with agents.

AI Agents are designed to plug into the Five9 Genius AI Suite, including AI Knowledge, Agent Assist and GenAI Studio. They can respond based on the context of the individual customer. For example, a customer inquiry about how many miles are needed to reach a certain loyalty status can be met with a response tailored to the mileage balance in the customer’s account.

Agents also automatically summarize customer interactions across voice and digital to give live agents context from previous interactions that can improve customer experience, such as upgrading a person to a new loyalty status without prompting.

Five9 said its AI approach gives businesses full control over the level of autonomy and the behavior of AI Agents. The autonomy level can be dialed up or down from “No Trust” to “High Trust,” and cover a wide range of conversations, including pre-scripted, dynamic and complex language constructs.

That’s important to preventing hallucinations, which can be catastrophic in customer engagement scenarios. Companies can apply retrieval-automatic generation to train  agents with internal data.

As trust increases, they can dial up the trust equation to handle intent recognition and manage multi-step conversations while maintaining tight control over fulfillment. At the highest trust levels, large language models can autonomously manage conversations and trigger actions through application program interfaces, a feature called agentic AI.

Five9 AI Agents will be available in beta test in the first quarter of 2025.

Image: SiliconANGLE/Ideogram