At NeuraFlash, we’ve seen a lot in the past 10 years in the Artificial Intelligence space for customer service. We’ve been a part of the growth and practical enterprise deployment of Natural Language Processing and Machine Learning applications, from the early days of Natural Language based IVR systems, to the emergence of Virtual Assistants and Chatbots. We now see AI assistants being deployed across multiple modalities and channels including mobile messaging, SMS, web chat, as well as traditional voice and e-mail. Early on, most of the AI applications for customer service were focused on “self-service”, i.e., helping customers find information, answer questions, and complete tasks without human intervention. Needless to say, some of these deployments were successful, and some were wildly not. In hindsight, a different approach to solving the same problems may have been more effective, and may have even accelerated the progress of A.I. solutions for customer service to this point. This different approach came about after we spent time with a company called DigitalGenius, pioneering a Human+AI™ approach to customer service – combining the best use of human and machine intelligence in a seamless way. That way, the agents work faster and more effectively to solve incoming requests and the A.I. does the grunt work – eliminating repetitive tasks like time-consuming case filling, classification and routing, prioritizing and even looking up the right answer.
If we only knew a sliver of what we know now back 10 years ago!
It’s pretty simple actually, and not a unique set of problems whatsoever. Since forever, humans have been inventing things that make tasks easier for humans, helping to complete them faster and more efficiently. A.I. based solutions for customer service are no different – make customer service representative’s (CSR) tasks easier, increase their efficiency, remove repetition, and allow CSRs to focus on more critical tasks. And while this is great for the CSR (more challenging work = increased job satisfaction), it’s also great for the business as each enhancement to the efficiency means reduced operational cost and the ability to improve customer service. And the customer experience is better as well!
The reason “self-service” AI applications emerged in the market first is because it’s the holy grail. If you can fully automate question answering, conversations and issue resolutions, then you can, theoretically, replace CSRs and provide faster sales/service at a very low cost. So unfortunately, early Artificial Intelligence applications started with this top down approach, trying to fully automate CSR interactions – the hardest problem of all. Unfortunately, the AI technology was not ready, the teams implementing the technology were not experienced enough to fully understand the best design patterns, etc., resulting in less than stellar solutions and experiences. Another problem emerged since these self-service applications did not include practical applications of Machine Learning. They needed to be manually optimized and tuned by humans in a very laborious and time consuming process – lengthening rollout timeframes for not only initial implementations, but each and every time changes needed to be made to models.
In hindsight, there may have been a better approach to solving the same problems that more closely models contact center teams, not just an individual CSR – like self-service apps try to accomplish.
In a contact center, CSRs work together and collaborate, share knowledge, and support one another when complex customer issues arise. When an individual CSR is presented with an issue or case that he or she is unfamiliar with, and the CSR isn’t sure how to resolve the problem, they escalate it to the team or to a supervisor for help. In the process, another specialized and more experienced CSR helps solve the issue, and the CSR assigned the issue learns through the process. Hopefully, the next time the same agent gets the same issue presented to them, they are equipped to resolve it without assistance by the team. So when introducing an Artificially Intelligent application into the contact center, wouldn’t it make enormous sense to have that AI app learn from CSRs, just like a newly hired team member would? If A.I. solutions for customer service started with this Human+AI™ approach a decade ago, rather than being somewhat autonomous applications trying to model CSRs and “deflect” inquiries from getting to the human CSR, would the capabilities of A.I. solutions for customer service be much further along? NeuraFlashsays yes.
AI+Human: CSRs Supporting Natural Language self-serviceAI applications should be viewed as members of the team, and even when they want to work autonomously, they will eventually need support. While the technologies and design strategies for conversational self-service across all channels has come a very long way, there is still much room for improvement. A number of self-service platforms (ChatBot for example) have recently started to implement capabilities for self-service apps to escalate the problem to human CSRs when they detect they can’t answer a specific question, or do not understand what it is being asked. Some platforms escalate the conversation to the CSR to fully take over and complete, and some do that for a specific trouble spot, but continue with the conversation once the human CSR helps get the conversation back on track for the self-service app to understand. In this scenario, the human is acting as a supervisor, helping the “AI CSR” to do its job the best it can.
Just as Human CSRs should be supporting the self-service AI applications with their tasks, AI applications should directly support Human CSRs do theirs. The NeuraFlash team recently spent time with a company called DigitalGenius, a leader and innovator in the area of Artificial Intelligence for customer service, to discuss Human+AI™ concepts. There are many tasks CSRs need to complete that are repetitive and less complex, but take significant amounts of time. That time could be better spent solving actual customer issues instead of filling case details, prioritizing or routing cases or even searching manually through answers repositories. AI applications using Machine Learning and Deep Learning can help CSRs by filling out forms and cases from customer inquiries, categorizing incidents and cases so the best CSR is routed each work item within the team, and providing recommended responses during customer interactions. What’s more, this approach helps CSR teams handle more cases as each company adopts social channels and develops more touch-points with its customers. DigitalGenius brings practical applications of deep learning and artificial intelligence into the customer service operations of large companies.
DigitalGenius has a unique approach to solving these problems and an impressive history. They apply new advances in Deep Learning to help transform customer service operations inside businesses. Customers can use the historical customer service transcripts to train the DigitalGenius A.I. and integrate it directly into the contact center’s existing software. Once enabled, DigitalGenius automates and increases the quality and efficiency of customer support conversations across text-based communication channels like email, chat, social media and mobile messaging.
DigitalGenius helps manage volume spikes during service disruptions, while opening new communication channels, or dealing with repetitive questions. By embedding a layer of Deep-Learning intelligence inside existing customer service operations, they enable large companies to unlock new communication channels and serve their customers more efficiently.
The trained neural networks works alongside the human agents, learning from them in real time and enabling them to spend less time doing repetitive work and more time solving genuine customer issues.
This company has been around for over 3 years now, having started out as an Einstein Bot and chatbot company for brands. They have seen how difficult it is to build and maintain a great bot that delivers spotless experience, so they decided to explore the Machine Learning and Deep Learning applications and never looked back since. DigitalGenius is not only language agnostic, but can also operate across multiple industries, with clients ranging from airline, consumer packaged goods to banking industries.
Recently, KLM Royal Dutch Airlines was named by Forbes to be “the first airline and probably the first enterprise worldwide to test how Artificial Intelligence could assist customer service agents in managing the increasing volume of interactions with customers over social media channels”. With the help of A.I., they handle over 100,000 social media customer service requests per week and the numbers are growing.
DigitalGenius and NeuraFlash are partnering to bring these powerful Machine Learning capabilities to enterprises across many industries.
Here’s what DigitalGenius’ team thought about the partnership: “We decided to partner with NeuraFlash because of their incredible domain expertise in the areas of A.I., Customer Service for the enterprise and Salesforce. They are the Salesforce Consulting Partner of choice, and are able to bring this unique combination to the table. Our partnership is a perfect fit of Innovative product from DigitalGenius and Customer Success, Implementation and ongoing A.I. optimization strategies from NeuraFlash.”
Needless to say, we at NeuraFlash are equally excited to continue implementing best-in-class A.I. products and Salesforce!