Have you made a call into a contact center lately and gotten a prompt like this, “I’d like to text you a link to guide you through (your interaction). Is that all right?”
This text-based response as part of a customer call isn’t restricted to any specific industry; the interaction might be a router return or an airline upgrade. Each customer interaction and resolution typically involves multiple steps and you could be talking to an automated agent or a live one. Whatever the case, when you agree to a link to guide you through the interaction, you are subsequently guided through what could be several screens to complete the interaction on your mobile phone. If you are a customer of United Airlines, chances are you were using conversational AI platform company NLX’s technology.
Earlier this month, NLX published a detailed case study of their United Airlines use case. Doing a little research to peel back the onion on what looked like interesting technology, I discovered that NLX had been part of the Enterprise Connect 2024 Innovation Showcase. In addition, I listened to Andrei Papancea, CEO, NLX, interviewed on an AWS GenAI Live! podcast, “Empowering next-generation AI systems: from infrastructure to implementation,” recorded on April 16th. Suitably intrigued, I talked to Papancea on April 18th to discuss some of my contact center-specific questions about NLX’s patented multimodal conversational AI technology.
NLX
Before discussing the latest NLX innovation, I’ll take one step back. The NLX platform was initially designed as an enterprise-grade conversational AI solution that allows organizations to build, deploy, and manage advanced chat and voice applications at scale. The company calls the NLX platform an “application layer for conversational AI.”
NLX Voice+ is a feature of the NLX conversational AI platform that enables multimodal customer experiences by synchronizing voice interactions with digital channels such as web, mobile, or IoT interfaces. The core idea is to enrich traditional voice-based customer service (like phone calls or IVR) with real-time, interactive visual elements, creating seamless and guided user journeys across multiple channels.
Several CCaaS vendors have built their applications on AWS. Most notable is Amazon Connect but Interactive Intelligence began building its microservices-based CCaaS, PureCloud (now Genesys Cloud CX), on AWS in 2014 and others followed, e.g., Cisco, NICE, Talkdesk and others. So when I asked Papancea why he chose to build NLX with the AWS platform, I predicted that Papancea’s answer would include considerations like ease of integration with existing contact center solutions in the market.
Instead, Papancea began his response by referencing his work at Columbia University, first as a computer science graduate student, and then as an adjunct professor. For real world hyperscaler-based assignments for students, he leaned into AWS: “I believe AWS has the most robust architecture and building blocks,” said Papancea. And AWS skills proved marketable. After his first semester teaching, over 60% of his students landed AWS internships – and his popularity as an instructor soared.
What are the AWS building blocks that Papancea referred to? “Services like EC2 (Amazon Elastic Compute Cloud), S3 (Amazon Simple Storage Service), DynamoDB (serverless NoSQL database service), and their security mechanisms are second to none,” Papancea responded. When it came to choosing a cloud infrastructure for his own startup in 2018, the choice was obvious.
AWS is also the infrastructure several CCaaS are built on. I asked Papancea how that has factored into the rollout and success of NLX. He replied that a sizable proportion of NLX’s contact center customers use Amazon Connect. (An architectural diagram of how NLX works with AWS services including Amazon Connect can be found here). Of those using another contact center solution, many are using Amazon Lex (the AWS chat builder), which also facilitates native integration with NLX.
NLX’s deep integration with AWS allows customers to easily migrate from one contact center and/or AI solution to another while maintaining a native ecosystem experience. According to Papancea, a customer using Genesys with Google Dialogflow could transition to Amazon Connect and Lex in just a few days, including quality assurance time. This is possible because NLX decouples the application layer from the underlying infrastructure, allowing customers to change their contact center platform without extensive re-engineering. The NLX platform is designed to be modular across different cloud and contact center environments, e.g., Genesys, Cisco, Amazon Connect and others.
“People always ask, what use cases do you solve?” Papancea replied that the right question is, “What problems are you trying to solve?” In 2022, when NLX and United began their collaboration, United was very clear about the problem they needed to solve. The post-lockdown surge in travel, combined with easing restrictions for cancelling and re-booking, resulted in millions more calls per year. Like many contact centers, United’s contact center was expected to manage the surge without dramatically increasing the number of agents. United Airlines started looking for ways to improve automation with voice calls.
At least for some customers, the answer to increased voice call automation was not a better voice bot but the addition of a digital channel. While contact centers often encourage callers to go to the website or mobile application to resolve their query, with NLX, the customer doesn’t have to hang up the phone and do something else. The voice assistant, powered by NLX Voice+, stays on the line with the customer, guiding them through inputting information digitally to complete their request – and the call does not drop. And, if the caller is not successful, the interaction can not only be transferred to a live agent but transferred with full context of what the customer has already successfully done.
“The most complex use cases, the ones that drive the highest ROI, are often too complicated for voice-only channels,” Papancea believes. On average, over the last six months, 31% of flight cancellation requests and 64% of wheelchair contact center requests at United Airlines have been automated through NLX. In the same period, customer satisfaction scores (CSAT) are 82% positive for cancellations and 90% for wheelchair assistance and handling requests. And NLX and United continue to solve problems together.
Many enterprise software companies – CRM and CCaaS vendors – are incorporating conversational AI directly into their platforms. I asked Papancea why enterprises choose NLX versus having a single stack with their CCaaS or CRM vendor. “There is AI infrastructure, there is cloud infrastructure, and there is contact center infrastructure,” Papancea answered. He joked that he didn’t believe that he would live long enough for all of those capabilities to be delivered in an all-in-one platform.
On a more serious level, Papancea said that the NLX platform is “unusually modular” for the space. Because AI is changing so rapidly, that modularity allows them to add and subtract new models or technology easily.