The Future of Work Chatbots, Voicebots, and Virtual Assistants

Enterprises increasingly rely on chatbots, voicebots, and virtual assistants to service more customers without adding to their staff count. These devices can be connected by a single underlying software platform, although there are several companies that use these devices independent of each other.

Future Markets Insights predicts that the chatbot market will grow from $696.4 million in 2023 to $4.9 billion in 2033, a 21.6% compound annual growth rate (CAGR), with a lower cost of operation and a greater emphasis on customer involvement through numerous channels contributing to the increase. The devices automate a variety of customer-facing and internal tasks, including answering customer queries, enabling customers to use self-service, providing a brief review of customer interactions, and summarizing financial and other reports for internal uses.

The voicebot market size is estimated to reach $98.2 billion by 2027, representing a CAGR of 18.6% during the forecast period 2022–2027, according to an IndustryARC report.

The report attributes the strong growth to the increasing adoption of smart home devices with built-in automatic speech recognition technology, along with the growing penetration of smart speakers and smartphones. Voicebots use machine learning, natural language processing, and interactive voice response navigation systems to answer customer queries by interpreting the intentions and meaning of speech.

Virtual assistants are used both by consumers and within enterprises. Statista estimates there will be more than 150 million voice assistants in the U.S. by 2026 as consumers and businesses continue to adopt new technologies to boost efficiency and convenience. Each one of these technologies has evolved noticeably in the last few years, and all are expected to advance even further as the embedded underlying AI, in many cases, generative AI (GenAI), continues to progress.

“Foundationally, I think AI is giving us an ability to step back and think about how customer experience organizations are approaching work and technology and rethinking their workflows, all of their processes, even each individual point, solution, and piece of technology,” said Elizabeth Tobey, NICE head of digital marketing and AI. She recommends tying together chatbots, voicebots, and virtual assistants via an underlying platform.

The AI in the technologies can range from very simple algorithms to the latest open source releases of different large language models (LLMs), with some using multiple AIs and LLMs. Others use proprietary AI or LLM technologies. “Many companies have been using some form of AI chatbot, especially on the customer service side, for many years, even prior to the emergence of generative AI,” added Don Schuerman, CTO and VP of marketing for Pegasystems.

“There are some websites where the chatbot is going to look like a chatbot from 2012,” said Frank Schneider, Verint VP and AI evangelist, “with some FAQs, and maybe a decision tree, if you’re lucky, which is more like 2014. And maybe, maybe, it plays the role of traffic cop to route calls to different agents or to offer limited self-service capabilities, which is probably like 2009.”

Most of the older devices can be upgraded to handle complex tasks relatively easily and quickly, according to Schneider. However, others are little more than slightly advanced IVRs that don’t lend themselves to easy upgrades to the latest technology.

With the hype around GenAI, enterprises with a circa 2019 chatbot want to add LLM capabilities to handle a modern environment, Schneider says. “They’re taking flat article content and using generative AI for summarization to augment conversational flows. One of the easiest things generative AI can do is take flat content and transform it into content you can converse with.”

Among other uses, this enables enterprises to crawl even very detailed websites for content to provide customers or agents assisting customers with interactions for very quick answers to complicated questions.

Many companies are still hesitant to use GenAI with chatbots and other technologies because the LLMs are still largely unpredictable, Schuerman said.

“If you ask it the same question twice, you might get two different answers. For customer- facing technology, you want a high level of predictability. You certainly don’t want the situation where a customer asks the same question twice and gets two different answers. So, what I’ve seen is a lot of organizations continue to use forms of AI that predate larger language models.”