Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.
From a smart assistant that helps you increase your credit card limit, to an airline chatbot that tells you if you can change your flight, to Alexa who operates your household appliances on command, conversational AI is everywhere in daily life. And now it is making its way into the enterprise.
Best understood as a combination of AI technologies — Natural Language Processing (NLP), Speech Recognition, and Deep Learning — conversation AI allows people and computers to have spoken or written conversations in everyday language in real-time. And, it is seeing good demand, with one source projecting that the market will grow 20% year on year to $32 billion by 2030.
Broader AI scope
Organizations have been quick to adopt conversational AI in front-end applications — for example, to answer routine service queries, support live call center agents with alerts and actionable insights, and personalize customer experiences. Now, they are also discovering its potential for deployment within internal enterprise systems and processes.
Popular enterprise use cases for conversational AI include the IT helpdesk where a bot can help employees resolve common problems with their laptops or business applications; human resource solutions for travel and expense reporting; and recruitment processes where a chatbot guides candidates through the company’s website or social media channel. It informs them on what documents they must submit and even makes preliminary selection of resumes.
MetaBeat will bring together thought leaders to give guidance on how metaverse technology will transform the way all industries communicate and do business on October 4 in San Francisco, CA.
While there is no denying that conversational AI offers attractive opportunities to innovate and differentiate, it presents some challenges, as well. Managing an enterprise conversational AI landscape with disparate technologies and solutions that do not communicate with each other is only one problem. Inadequate automation of repetitive processes across the conversational AI lifecycle and the lack of an integrated development approach can extend the implementation timeline. Last but by no means least, AI talent is in short supply.
By adopting some thoughtful practices, enterprises can improve their conversational AI outcomes.
Five best practices for successful conversational AI
1. Do it with purpose
Conversational AI should be implemented with a specific purpose, and not just as a gimmick. Questions, such as what kind of experience to provide to customers, employees, and partners, and how to align conversational AI with organizational goals, will help to identify the right purpose. Also, the solution should address activities involving the processing of multiple data points — for example, answering questions about loan eligibility, which can add significant value to the customer experience — rather than working on tasks that can be accomplished with predefined shortcuts.
2. Mind your language
Taking a conversation-first approach is important for scaling technology across the enterprise. But since different people speak naturally in different ways, the understanding must extend not only to the words being used but also the intent. If the NLP solution being used is not capable enough, it will create friction in the interaction.
3. Do it yourself
Low-code/no-code platforms are giving rise to citizen developers, that is, business or non-technical employees who write software applications without the involvement of IT staff. Going forward, this could help to overcome the shortage of AI skills plaguing most enterprises.
4. Personalize, extremely
Among the many features of conversational AI are contextual awareness and intent recognition. The technology can recall and translate massive information from past conversations in human-like fashion, and also understand what the speakers are asking even when they don’t “follow the script.” These capabilities yield remembered insights that enterprises can exploit to personalize everything to individual preferences, from products and services to offers and experiences.
5. Eye on the past and the future
Conversational AI should take an approach that relies on historical insights and continuous post-production evolution using telemetry data on user demands, to improve stickiness and adoption. Strategically speaking, organizations must incorporate good governance when automating a conversational AI lifecycle. This means that, irrespective of the technology being used, the underlying architecture must support plug-and-play and the organization should be able to benefit from using the new technology.
In short, to gain traction within the enterprise, conversational AI should enable intelligent, convenient, and informed decisions at any point in the user journey. A holistic and technology-agnostic approach, good governance, and internal lifecycle automation with supportive development operations are the key factors of success in conversational AI implementation.
Bali (Balakrishna) DR is senior vice president, service offering head — ECS, AI and Automation at Infosys.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!
Author: Balakrishna DR, Infosys