)
As part of our ongoing “Life at Uniphore” series, we interviewed Balaji Raghavan, Uniphore’s new Chief Technology Officer. Balaji shared how his background with Google, Lyft and other tech pioneers has shaped the customer-first vision he brings to Uniphore. He also broke down the essential pieces for conversational AI and automation success and revealed how today’s technology is reshaping agent coaching in real-time.
Uniphore: Tell us about your role at Uniphore and your background.
Balaji Raghavan: As CTO at Uniphore, I am responsible for the technology choices in AI solutions and the services and platforms that deliver integrated experiences to the users. Uniphore is at the forefront of conversational automation and emotion intelligence, which got me excited to join this journey to improve our experiences in everyday conversations. Conversations are a complex domain with many nuances, like how a person would react in person versus in, say, comments on Twitter. Similarly, it takes a while to learn whether our audience is showing interest in what we are talking about. So, the ability to get more context about the person, their body language, the tone and sentiment in their voice, or just their interest and engagement in the conversation—while also giving us hints to say things in a more empathetic way—would make us more effective and efficient in our conversations.
Throughout my career, I have worked on products that put the user first and work towards making the world around us better and more efficient. From leading Gmail at Google for a decade to deliver the most efficient way to communicate on the web to founding FinitePaths to build a Q&A platform to provide the best answers to customers for their everyday questions to then leading the Rideshare Marketplace at Lyft to build the best transportation platform to reduce traffic and take steps towards replacing roads with parks, I have been fortunate to work on finding ways to use technology to solve the inefficiencies in our lives that are taking time away from important things like spending time with our loved ones.
“Uniphore is at the forefront of conversational automation and emotion intelligence, which got me excited to join this journey”
U: What are the building blocks of a successful Conversational AI and automation platform?
BR: Conversations happen in different forms – text, voice, video. A conversational AI and automation platform needs to be able to work with any or all of these modes. It can be broken down into a few AI component technologies and the interfaces to normalize and utilize them. The main AI technologies that drive this experience are:
Speech recognition: Speech recognition allows us to recognize every word, the language it maps to and the right attributes when paired with tonal and sentiment analysis. [This allows us] to convert speech into a format that can be utilized by other components (e.g. as a transcript or in a data structure like a linked list).
Language understanding: It is important to identify various parts of text/speech to find the intents (what this sentence is about - e.g. buying a car) and entities (what specific objects are being described - e.g. the make and model of the car). These intents and entities form the context of the discussion and become the building blocks for being able to assist the conversation.
Image recognition: For conversations that happen over video, we need to find the right participants at all times and understand their sentiment and engagement to find the best ways to guide the speaker to reengage the audience or to introspect in real-time on what could have triggered a negative sentiment. Conversations over video can be dual-party or could be multi-party where not everyone is visible in a video conferencing call; but the AI can see all the participants and provide helpful summaries of these visual cues to us.
With some or all these building blocks, there are two components that help Uniphore provide an effective automation platform to our customers:
Flow designer: Every transactional conversation has an ideal flow to achieve best outcomes. For most businesses, it is important to have their employees learn and follow these flows to a tee to improve their chances of success. We provide the AI tools that analyze and transform inputs for these flows; a flow designer helps the customers use their deep understanding of their business to find the optimal workflows to drive customer or employee satisfaction.
Robotic Process Automation: Similarly, in transactional conversations, customers are trying to get some action to occur in the transaction which tends to be a core flow in a business, like canceling an account. Based on custom rules, this can be executed in a way where several tools, like CRM, can all be updated using Robotic Process Automation.
Additionally, we need to have the ability to work with several input providers, like different SBCs for voice for each of these modes of communication as a SaaS (Software as a Service) business, so our customers don’t need to change their current infrastructure and data to work with our platform. These integrations enable our customers to adopt our product offerings and get value from it without having to re-architect their products.
U: What does customer experience mean to you?
BR: For me, customer experience means what it reads (i.e., it is the experience of the customer with the product). When building products, the goal is always to improve the customer experience; but for most companies, the experience of the customer extends beyond just the product they sell (e.g., email/phone/SMS marketing, frictionless checkout, customer support, etc.). These are common needs for most companies and opportunities for companies like ours to enable other businesses to focus on their core products instead of these experiences.
U: How do you see AI and automation affecting real-time agent coaching?
BR: AI and automation are crucial for agents to learn and adapt quickly to deliver positive experiences for the customer. Traditionally, agent coaching depended on supervisors replaying a sample of the recorded calls to find areas of improvement. AI can find patterns from similar calls where the agents were able to provide the best customer experience and make those available to the agent to coach them on achieving similar results. AI can also make the appropriate information about the customer, documents and tools available for the agent to make decisions based on the intent of the call. That way, the agent does not need to spend time finding and reading the proper materials to assist the customer. As a result, we can achieve better First Call Resolution (FCR) and reduce Average Handling Time (AHT) to achieve efficiency for both the agent and the customer. This makes even the novice agents seem very knowledgeable about the business and deliver quality service.
“AI and automation are crucial for agents to learn and adapt quickly to deliver positive experiences for the customer”
U: What are the top 3 innovations in AI that you see in 5 to 10 years out?
BR: There will be several innovations, and “top” is a subjective choice. The ones that I am looking forward to are:
- Use of AI in the medical field to aid doctors through use of the data from other doctor experiences across the world and the knowledge of the patient’s genome to diagnose their condition. This can also help with prognosis of future ailments and provide ways to mitigate them early.
- Use of AI to provide independence to people with special needs, like those with disabilities or elderly people or even young kids, to perform common tasks, like reaching for objects safely or opening a jar carefully, or more complex tasks, like mowing a lawn or fixing leaks in pipes.
- Use of AI in transportation to completely replace all people-owned cars with vehicles that arrive on demand and can communicate with each other to avoid accidents in a hub-and-spoke road network. I still believe in the mission that both my previous employers – Google and Lyft – are building towards.
Fun rapid-fire questions:
U: Favorite sport to play.
BR: I love playing badminton, table tennis and basketball.
U: Favorite sport to watch.
BR: I prefer playing sports to watching them. I do not follow any sport currently but enjoy watching football (soccer for my friends in the U.S.), tennis, gymnastics, swimming and athletics.
U: Favorite restaurant to get takeout from.
BR: Dunkin’s Donuts! Chocolate is my weakness, and a Double Chocolate Donut could easily become a daily breakfast if I could enjoy the taste without the ill effects of eating them.
U: First word that comes to mind when you think of AI.
BR: “Enablement” – AI helps make mundane things easier and removes the need for attention to immaterial details. At the same time, it delivers consistent attention to details that we might miss and makes those experiences efficient and precise (e.g., if every driver was equally proficient at driving and equally attentive on the road, the chances of error/accidents reduce drastically).
U: Last book you read or audio book listened.
BR: The last book I read (or re-read to be precise) was Working Backwards by Colin Bryar and Bill Carr. There are so many interesting details of methods to apply in your leadership and execution on projects that it takes a few reads to adapt the lessons in your teams.