By Tom Haunert
Chatbots replace conventional application user interfaces with human conversation. Chatbots combine and connect messaging channels, artificial intelligence (AI), and back-end integrations to help enterprises automate these conversations.
Oracle Magazine sat down with Suhas Uliyar, vice president of mobile, bot, and AI strategy at Oracle, to talk about chatbot and AI technologies, chatbot challenges, and chatbot solutions from Oracle.
Oracle Magazine: What are chatbots, and why are they important?
Uliyar: Simply defined, a chatbot is a computer program designed to simulate conversation with human users.
Chatbots are important because they acknowledge and address a change in engagement preferences. The last decade saw a major adoption of mobile as an engagement channel for consumers and employees within the enterprise. But now we’re seeing an increase in the adoption and use of messaging channels including Facebook Messenger, WhatsApp, WeChat, Slack, and SMS, and voice personal assistants including Amazon Echo Dot, Google Home, ApplePod, and so on as preferred engagement channels.
Messaging channel adoption is happening quickly because it is instant, available 24/7, and people can use natural language and get a consistent experience across multiple devices. This is leading to innovations and use cases for chatbots powered by artificial intelligence that can help enterprises automate conversations at scale through these messaging and voice channels.
Oracle Magazine: How do emerging technologies—including AI, machine learning, and natural language processing—support chatbots, and what kind of challenges do they introduce to chatbot development?
Uliyar: One of the primary reasons for chatbots is to enable end users to communicate with natural language, so the first aspect of the AI and machine learning technology support is all about being able to understand the end-user conversations. A primary component of this AI technology is natural language understanding [NLU] and natural language processing [NLP].
NLP applications attempt to understand natural human communication, either written or spoken, and communicate with us using similar, natural language. Machine learning helps machines understand the vast nuances in human language and to learn to respond in a way that an audience is likely to comprehend.
There are many reasons why AI technologies are viable and important right now. The two main ones are that, first, compute power has increased quite a lot and can handle the sophistication of the machine learning algorithms behind AI technologies. And second, the increase in access to data over the last decade means that machine learning has more to learn from, so it can be more accurate and more deterministic.
The challenges that come with AI machine learning include having enough data to learn from and understanding where and how to start developing projects that include chatbots, AI, machine learning, and other emerging technologies. At Oracle, we’ve been focusing on helping customers drive toward a much more effective conversational AI platform that can help them get over the hump of implementing chatbots.
Oracle Magazine: What are the biggest challenges for companies that want to chatbot-enable existing apps or develop new chatbot apps?
Uliyar: Our customers want to know where and how to get started with chatbots or conversational AI. And that means defining and fine-tuning the use cases for the chatbot. Organizations can look at what they’ve implemented in their mobile apps as a quick start toward bots. They can use their apps to look at the typical repeated questions from customers and the standard responses.
After defining use cases, the next challenge to getting started with chatbot projects is determining what channels to support. There are a plethora of different channels such as Facebook Messenger, WhatsApp, WeChat, Line, Telegram, Skype, Alexa, Amazon Echo Dot, Google Home, and more.
The next challenge is to then determine the scope of a bot as it relates to what to do when the bot is unable to answer questions, either because the bot is not capable of answering those questions or the bot hasn’t been configured to answer the questions that the end users are asking. If a chatbot doesn’t cover all of the organization’s use cases or something is not configured properly, how does your chatbot still provide the best possible user experience?
And finally, how does the organization integrate the bot into their enterprise? How do they securely connect the chatbot to multiple messaging channels and then connect to the different systems of record in an effective way to provide timely responses and the best possible user experience?
Oracle Magazine: What are Oracle’s chatbot technology solutions, and how do they address enterprise and developer challenges for building chatbots?
Uliyar: We developed a mobile service—Oracle Mobile Cloud Service—in Oracle Cloud that has widespread global customer adoption across industries. We’ve expanded our offering and built our Oracle Intelligent Bots platform on top of the mobile service to offer Oracle Mobile Cloud Enterprise. This new platform is powered by some emerging technologies, including AI that uses machine learning algorithms, to help provide a high level of intelligent engagement with end users.
The Oracle Intelligent Bots platform addresses customer challenges in several ways. First, it’s a comprehensive end-to-end-platform with everything an enterprise needs to deliver a successful chatbot solution to its customers. The bot platform provides an abstraction layer to integrate with the different channels, including Facebook Messenger, WeChat, WhatsApp, Alexa, Google, and so on.
The second benefit for the enterprise and developers is that we provide the full NLU service with our dialogue engine and a tool that helps the customer define and refine use cases. The NLU engine is implemented as a pipeline of algorithms that can help customers train the model based on the amount of data they have. Customers that have limited or no data can just provide a few sample phrases to get the model started, and as the system gets used, the algorithms auto-adjust to learn from the expanded dataset. This provides organizations and developers with a design principle and the tooling to help them understand how to model the natural language very simply, because a big challenge again for some customers is that they don’t have enough data to support all of their use cases.
And at the same time, we support development that lets a bot pass on new queries with full context to a human agent in case the bot either cannot answer the question because it hasn’t been programmed to or it just needs to be configured correctly. So, the human agent handoff and bot design that can clearly articulate what it can and cannot do are design principles that we also enabled as part of our dialogue engine and our NLU engine.
The third benefit for the enterprise and developers is the conversational AI capabilities that are integrated with the Oracle Intelligent Bots platform and provide all the machine learning algorithms, the cognitive services, the dialogue and context services, knowledge services, and data and insights for reporting.
Finally, the Oracle Intelligent Bots platform addresses the challenges that customers face when they need to integrate back-end systems with their bots. The bot platform includes a full integration stack that leverages current system integrations and investments in integrating with back-end systems so bots can more easily include a comprehensive end-to-end flow, enterprise security, release management capabilities, version control, and so on over the lifecycle of a bot.
LEARN more about Oracle Cloud for mobile and chatbots.
READ more about Oracle Mobile Cloud Enterprise.
Photography by Phil Saltonstall