Can you explain how conversational AI helps to create high-quality, consistent call experiences at scale for enterprise organizations?
Flexibility is the biggest advantage conversational AI has over conventional chatbot solutions. Typical rule-based chatbots are incredibly limited in scope and offer no genuine artificial intelligence, relying on buttons to drive conversations forward. Conversational AI, on the other hand, uses natural language technologies that accurately reflect the way humans interact with one another. Language understanding is a huge part of what sets conversational AI apart, as is being able to manage large volumes of chat traffic at scale effectively. Many solutions on the market today claim to be able to handle between 100-200 general topic intents. If you consider the potentially thousands of tricky policy questions that a large-scale insurance company might field from their customers, this actually doesn’t amount to much.
Boost.ai’s solution focuses on offering pre-built, industry-specific content, across a number of key verticals, with thousands of intents, not just a few hundred. Future-proofing a virtual agent with a foundation of industry-specific intents is crucial. A business can then layer their own brand’s unique content on top to build the advanced customer interactions that their customers will demand in the future.
What would you say is the key to ensuring effective human-machine interaction?
Even in human conversations, people can misunderstand each other and often use linguistic repair mechanisms to get a conversation back on track. Our solution contains a proprietary algorithm called Automatic Semantic Understanding (ASU) that essentially does that same thing, allowing the conversational AI to make smarter decisions about what it can and can’t answer. ASU helps to interpret whether it should handover customers to the correct human support staff if it identifies a request outside of its defined scope.
We recently conducted research that found that ASU replies more closely resemble these natural conversational repair mechanisms. The research showed that when the virtual agent expresses uncertainty and suggests a likely alternative, it doesn’t deter customers from reaching a successful outcome in the same way that a simple “I don’t understand” response might. This suggests that the best customer experiences are those that give consumers the answers they want, whether it’s from a human, a machine, or both.
Can you share an example of a successful implementation of Boost.ai’s conversational AI?
One of our biggest successes to date has been with Norway’s largest bank, DNB. We launched their virtual agent, AINO, in June of 2018, intending to reduce the high volume of online chat traffic that they receive daily. At the time, they were averaging several thousand interactions per day, which was overwhelming their support staff. We worked closely with DNB to develop their virtual agent and to implement a ‘chat-first’ strategy. That partnership grows stronger every day.
All support inquiries via the bank’s website were routed via the virtual agent first, and this resulted in an automation rate of over 50% in the first six months. Fast forward to 2020 and AINO now accounts for 20% of all customer service inquiries for DNB, across all channels. We are seeing similar success with not only our other banking clients but also the insurance and telecommunications companies that we work with here in Europe.
What role will humans play in customer service in the future?
There are two specific paths that we are already seeing the typical customer service role evolve into the adoption of conversational AI.
Firstly, by having a virtual agent automate the bulk of a company’s menial, repetitive inquiries; it gives existing support staff more time to focus on trickier customer cases. Benefitting the brand because its customers don’t feel like they are being rushed off the phone to clear a queue.
The other change (and this is more specific to our solution) is leveraging the skills and expertise of existing support staff to help train and maintain a company’s virtual agent. Our technology is relatively straightforward to use, and existing support staff can be upskilled into what we call ‘AI Trainers.’
To date, we have certified over 1,500 AI trainers for our clients, many of whom have come from existing roles within their organizations. Their extensive product knowledge doesn’t go to waste and, instead, is used to improve the digital customer experience from behind the scenes.
What role does analytics have in Boost.ai’s solution? Can users measure and improve their customer service function using your tech?
We are laser-focused on the enterprise space and have designed the Boost.ai platform with the needs of large organizations very much in mind. Analytics are a huge part of this. We offer a deep dive into the interactions of our virtual agents with the ability to monitor conversation ‘goals’ towards a specific business outcome. This helps to see what is and isn’t working a particular conversation flow.
Our next update, which is due to roll out in early 2020, will evolve Boost.ai’s conversational AI data analysis abilities even further. Instead of relying solely on AI trainers to identify missing intents, our algorithms will be able to make their own recommendations on which topics and intents are missing from a client’s model, purely based on an analysis of conversation data. This will be a game-changer in helping to improve customer experience in a more efficient and data-driven way.
“Language understanding is a huge part of what sets conversational AI apart, as is being able to manage large volumes of chat traffic at scale effectively.”
Lars Selsås, CEO of Boost.ai
Lars Selsås is the CEO of Boost.ai – Lars is the technical powerhouse behind Boost.ai, and his groundbreaking code is the basis on which the company’s conversational AI is founded. He has a deep understanding of high-performance code, big data, deep learning, and natural language technology. Lars spent several years in Silicon Valley working for some of the world’s leading artificial intelligence and natural language technology companies before returning to Norway to found Boost.ai.