Sramana Mitra: When you are implementing one of these use cases, could you double-click down on the other systems that you are integrating with? To make your system effective, you have to integrate with other systems. Walk us through one or two of those integration scenarios.
Raj Koneru: The virtual assistant is deployed on a company’s website or in their mobile app. Sometimes, it’s also deployed on email, SMS, or social messaging apps.
The functionality of the virtual assistant itself may require integration to the customer’s backend systems, which we enable the customer to do either by calling standard web service APIs or by using a bot in the backend. We’ve integrated with all the RPA players. That’s on the digital side.
On the voice side, we are pre-integrated with systems that are built on our platform and deploy it in Amazon Alexa or Google Assistant. We’re also integrated with all the IVRs in the market like Cisco and Avaya.
One would make a phone call to an IVR. That gets converted to text, the text is handled by the bot, and the bot will reply back in voice to automate the phone call.
Sramana Mitra: One of the key functionalities to be able to do this kind of automated customer response management is a knowledge base. What is your approach towards knowledge bases?
Raj Koneru: There are two components that you have in a virtual assistant – the ability to answer questions and the ability to perform a transaction. To answer questions, we have something called a Knowledge AI Framework, which enables the customer to extract FAQs and answers from unstructured documents like PDF and Word, as well as from websites.
It creates what’s called a knowledge collection. Then the customer organizes that knowledge. For the transactional side, we have something we call a Dialog Builder, which allows the customer to create a dialog flow, which then calls APIs or RPA bots to access backend data. It engages the customer in a conversation.
During the conversation, the context of that is maintained so you can provide a personalized response. That’s as far as the customer experience is concerned. Of course, we can make an API call to an existing knowledge base and get the results back.
Sramana Mitra: You said you go to market as a platform player and you distinguish yourself from other specific solution providers that focus on specific use cases. Does that mean that you have developers developing on top of your platform?
Raj Koneru: We go to market both as a platform and as a solution player. The platform is one thing. We also have full prebuilt solutions.
One is called Smart Assist, which is a call center automation solution. It enables our customer to transfer a call to us in the cloud or on premise. We provide an application that enables the customer to define the experience that call should have. Then we bring in a virtual assistant to automate the call. Then we deflect that from voice to chat optionality.
In chat, we automate that using the same virtual assistant and finally transfer it to a human chat agent. They integrate to 12 different live chat software. That’s for call center automation.
Then we have a product called Findly, which is a combination of search and chat. It produces a conversational search experience. It transforms the current search experience that you see on websites and mobile apps into a more conversational experience.
You can index a lot of knowledge and content. You can train that content. You can create rules. The chatbot is also attached to Findly. Depending on what the user may say to the assistant, it will either produce a search result or put you into chat to conduct a transaction.
There’s a third solution which is called Kora, which is the enterprise virtual assistant. Within Kora, we have the IT support assistant, HR support assistant, and the productivity assistant for those use cases, which are prebuilt solutions for each of those use cases in which you can then extend the underlying platform.
The fourth solution is BankAssist, which has all of the common banking functions like getting your balances, looking up your transactions, and transferring money. These are pre-trained models with predefined flows.
These four solutions get you 50% to 90% of the way depending on your requirements. The remainder can be accomplished with the underlying platform. We are unique in that we provide the solution and the underlying platform as opposed to many of the solution providers. Many of the solution providers don’t have the underlying platform or don’t have the front end solutions.