How Voiceflow’s Customer Service Generative AI Agent Tico Halved Support Costs
Conversational AI design and development startup Voiceflow has released a new report demonstrating how it used generative AI to successfully enhance its own customer support operations. Voiceflow deployed the generative AI-powered support agent Tico about three months ago and may use it as a template for the future of AI customer service platforms after it boosted efficiency and cut costs significantly for the startup.
Voiceflow developed Tico on its proprietary conversational AI platform with the goal of providing prompt and accurate responses to user queries without sacrificing support quality. Key to Tico’s functionality is its integration with a comprehensive knowledge base that encompasses Voiceflow documentation, support articles, and inputs from the community. Utilizing large language models like GPT-4, Tico can understand and address complex support queries by drawing relevant information from this extensive repository.
“Tico was born out of a necessity to redefine customer experience and support at Voiceflow. Being an AI agent-building platform, we wanted to reimagine how modern customer support could be offered through the art of the possible that Voiceflow offers. From an operational lens, for us to win, our users need to win,” Voiceflow head of customer experience and AI assistants, Tahsim Ahmed, told Voicebot in an email interview. “The goal was to create a system that not only provided immediate, accurate assistance to users but also embodied the innovative spirit of Voiceflow. Tico is a testament to our commitment to leveraging conversational AI to not just augment but revolutionize the way enterprises handle customer interactions.”
Voiceflow designed Tico to emulate its brand voice and then embedded it into all of its existing support channels about three months ago. Since its deployment, Tico has achieved a nearly perfect rate of response. Tico has a 97% deflection rate in handling support inquiries, significantly reducing the workload on human support agents. The AI assistant also successfully resolved 90.9% of support tickets, maintaining a high customer satisfaction score of 93%.
Since launch, Tico has processed over 1,300 support tickets, a volume typically managed by a human agent in a quarter but at a fraction of the cost. The implementation of Tico has allowed Voiceflow’s human support staff to concentrate on more complex and nuanced issues, requiring a personalized approach. Remarkably, while handling 64% of all inquiries, Tico accounted for only 4% of the total support operational expenses.
This increased efficiency also resulted in significantly faster response times compared to traditional support channels. The integration with the knowledge base further reduced the need for human intervention in support cases from 16.1% to 2.5%. Voiceflow attributes the rise in overall customer satisfaction and the liberation of staff resources for product development and innovation to Tico. The company’s investment in Tico turned profitable within three months, with the AI assistant generating substantial net savings and cutting direct support costs by 48%.
“The integration of the knowledge base was a game-changer, and Tico’s single most important pillar,” Ahmed said. “It transformed Tico, and our entire Product Support offering, from a simple FAQ intent-entity querying-answering bot, to an intelligent agent capable of understanding context and delivering nuanced, specific and tailored responses.”
The slight improvement in customer satisfaction levels to 93.8% post-Tico’s introduction underscores the project’s success. Voiceflow views Tico as the “catalyst that drives our vision of democratizing AI-driven communication,” according to Ahmed. It can serve as a model for how conversational AI can revolutionize customer service, efficiently addressing routine questions and basic issues while avoiding the quality drop often associated with full automation.
Looking ahead, Voiceflow plans ongoing enhancements to Tico’s capabilities, potentially further minimizing the necessity for human support and facilitating more personalized, context-aware responses. This case study serves as a compelling example of how AI can empower customer service teams. Rather than completely replacing human agents, AI tools like Tico can take on repetitive tasks, enabling human agents to focus on providing high-quality, person-to-person service. Voiceflow aimed the $15 million it raised in August toward generative AI plans. Ahmed predicted that Tico’s success in a short time will encourage further developments in the same vein.
“With Tico, we demonstrated how AI can elevate customer support from a reactive, transactional role to a dynamic, insightful part of the business fabric, influencing product development, marketing strategies, and overall customer engagement. This is the kind of holistic, integrated approach that will define the leading edge of customer experience, operations and product support in 2024 and beyond,” Ahmed said. “For our customers, Tico serves as the industry answer to the legacy-problem of effective Customer Support at scale with Generative AI, but more importantly a beacon of powerful workflow & task-driven AI agents on Voiceflow,” Ahmed said. “Voiceflow AI Agents, like Tico, underscore the potential of Voiceflow’s platform, demonstrating how our users can build their own sophisticated AI agents.”