Tell us about your role and the team/technology you handle at boost.ai.
I work as Chief Data Scientist at boost.ai, a Norwegian software company specializing in conversational Artificial Intelligence. My role involves developing state-of-the-art algorithms for Natural Language processing and Natural Language understanding, as well as managing the Machine Learning team. I’m responsible for the machine work behind-the-scenes and building the backbone of our solution. Our work ensures that any customers interacting with a business (be it a bank, insurance company or telco) via one of our virtual agents get served correct and useful replies. We develop algorithms ranging from language detection, entity extraction, intent prediction, and our proprietary automatic semantic understanding.
Why are B2B marketing teams steadily moving toward Applied Data Science for Sales and Marketing initiatives?
The more data that a Marketing team has at their disposal, the better they can accomplish their goals. Artificial Intelligence is a perfect platform for helping both gather and parse the mountain of data that marketers need in order to effectively do their jobs. Virtual agents are one such avenue that can be utilized to achieve this. We recommend our clients position their Virtual agents as a ‘front-line’ service channel – this not only gives their customers 24/7 access to the answers they need, but it also helps gathers useful customer data that can be utilized by sales and marketing teams to better target their efforts.
How do you build Analytics around Conversational AI platforms?
Our analytics dashboard offers a deep dive into the interactions of a Virtual agent. Apart from being able to track things like a number of conversations or how many buttons and links are actioned on, we also monitor message and conversation feedback. This feature is popular amongst our clients as it allows them to see how certain intents and responses track with their customers. We have also recently implemented a ‘goals’ system that monitors conversations towards specific business goals. This is a great way to increase conversion rates and see what is and isn’t working in a particular conversation flow.
What is conversational AI? What types of technologies run this engine?
Conversational AI is the technology powering the next generation of what some people call chatbots. At boost.ai, we think they’re better than that and call them Virtual agents. Using a combination of Deep Learning and Natural Language understanding – as well as a number of our proprietary neural networks – our conversational AI is able to better understand the direction that a conversation is headed and offer intelligent and helpful responses with far greater accuracy than in other solutions. In fact, with our latest Automatic Semantic Understanding (ASU) upgrade, Virtual agents built on our technology are able to reduce false positives in complex customer interactions by up to 90 percent.
How can data analytics make virtual agents function more effectively?
Data Analytics opens up a world of possibilities for maintaining and improving Conversational AI. For example, if a bank identifies that its customers are regularly asking about a particular product in its portfolio, it can choose to have its Virtual agent proactively mention that product either the next time the customer returns to the bank’s website, or mention it to other customers who might be interested. Similarly, Analytics can be used to find where the gaps in the model might be and generate new Intents and responses almost on-the-fly so the Virtual agent remains constantly up-to-date.
Tell us about your Customer Experience products. How can MarTech customers benefit from your AI/ML products?
Customer Experience is a primary focus of boost.ai. We believe that if you put Customer Experience first, the rest will follow. Which is why our solution has been developed with a CX mindset from the ground up. While our market-leading algorithms are unparalleled in their ability to understand the underlying Intent and context of the customer interaction, we’ve also developed them to know when it’s appropriate to transfer to a human rather than just answer with a canned ‘I don’t understand’ response. This approach to CX has proved immensely successful for our clients with one major Norwegian bank, DNB, managing in just six months to automate 51% of all chat traffic via their Virtual agent.
What is the current state of Machine Learning and AI for Digital Transformation?
Machine learning and AI are the catalysts for Digital Transformation. Adoption of these technologies by industries is constantly showing significant results. It is more effective and more accurate decision-making in real-time and can help to dramatically accelerate a business. Every major industry is exploring and educating their employees to approach challenges using data-driven decisions.
Can you recommend some best practices when building out a conversation AI solution?
At boost.ai, we maintain that it is important that a Virtual agent can be developed and maintained by non-technical personnel. Finding a vendor that provides extensive training material and online certification means you can upskill existing customer support staff into AI trainers. This is key, as it is their expertise and product knowledge that can prove invaluable to maintaining and further developing a conversational AI project after launch.
Abhishek Thakur serves as Chief Data Scientist at boost.ai. After graduating with a Master’s degree in Computer Science from Bonn University in 2014, Abhishek has been working with automatic Machine Learning. He is a former rank 3 in Kaggle competitions.
boost.ai provides solutions to Scandinavian Banking, Finance, Retail, Transportation, Government, and Insurance – solutions that make headlines while boost.ai grows at a record rate.
Now the company employs 40 ambitious individuals with the same goal: To deliver virtual assistants to all industries, enhancing all departments.