While Google, Amazon, Apple, and Nuance remain the most well-known names, other companies have quietly built natural language/conversational AI businesses. Consider China’s iFLYTEK, which reported more than $6 billion in revenue for its last fiscal year on solutions such as its aftermarket car assistant. And Boston area-based Interactions might be the most successful customer service virtual assistant company on the planet.
A Focus on Customer Service Automation
Founded in 2004, Interactions has been focused on customer service automation, providing a natural language-powered virtual assistant for brands such as Hyatt. Throughout Tractica’s virtual assistant research, we have maintained that the most lucrative use case for enterprise virtual assistants is customer service and marketing. However, some larger firms with conversational AI capabilities have shied away from the customer service use case, most likely because many enterprises consider virtual assistants to be too risky of a proposition as a reliable first line engagement for their customers. Instead, these conversational AI firms have looked to automating clinical documentation in healthcare or automating internal IT operations. Interactions is one of just a few conversational AI companies worldwide that has not deviated from a primary focus on customer service automation.
This determination is starting to pay off. First mover companies such as Hyatt and others have embraced Interactions’ virtual assistant approach. Since March 2015, Interactions has handled more than 2 billion calls for its customers and processed 5.5 billion utterances. The company reported that it has booked $100 million in revenue, though it noted that this is not an annual revenue figure yet.
Evidence of AI Scale
At Interactions’ recent Analyst Day, CEO Mike Iacobucci said the company, which acquired AT&T Labs’ Watson speech recognition and natural language platform and team in December 2014, is focused on scaling its solution. And it appears that Interactions is making progress. According to senior vice president of Operations Michael Roche, the company shows evidence of AI scale.
Roche explained that as the platform processes calls, the company uses what it calls intent analysts to help identify caller intent to guide the AI. Intent analysts come in only when AI gets stuck. In 2016, Interactions employed 1,500 intent analysts. Today, the company uses the same number of intent analysts – even with exponential growth in transaction volume – because Interactions’ AI is getting better. Roche estimates that if the AI had not improved, Interactions would require 5,200 intent analysts to handle the current transaction volume.
New Opportunities and a Shared Risk Approach
Interactions’ successful platform is opening new opportunities for the company because it can point to solid ROI results. Some of Interactions’ estimates show it can cut call center costs by 50%, but that the percentage depends on the customer. The company said some customers are saving money and use Interactions’ automation instead of hiring offshore resources, which allows them to use a smaller core of domestic-based customer service agents. Perhaps key to convincing potential Interactions customers to come on board is the company’s shared risk approach: success-based pricing. The company gets paid by reaching transactional goals that are set by Interactions and its customers. For example, Interactions is paid if Interactions’ virtual assistant successfully completes a caller’s request (what is my balance, when is checkout, etc.).
Conversational AI is not easy. There are very few companies today that have the competency to succeed in providing solid solutions. But more conversational AI companies are jumping into the customer service market. Look for many to follow Interactions’ shared risk approach to open the market.