Times have changed in the startup world. Just a few years ago, nearly every hot new startup was a mobile app, a peer-to-peer service, a social media app or e-commerce play. Customer service and voice technology, on the other hand, were seen as dull and unglamorous – the domain of dusty old call centers and automated phone systems.

The rise of Generative AI (GenAI) and cloud computing breathed new life into the moribund contact center ecosystem. Today, founders are recognizing the vast potential of large language models (LLMs) to revolutionize the way businesses interact with their customers, and investors seem to be taking notice.

Contact Center Software Having Its Moment

I recently stumbled upon this video showcasing the winners of a 2024 UC Berkeley’s AI Hackathon, and what caught my attention was the number of startups focusing on customer service applications. Two out of the eight presenting companies had developed solutions targeting this space. One of them, Spark, demonstrated an agent performance solution that used AI to analyze sales pitches and provide coaching for improvement. The team showcased a use case where entrepreneurs could upload a recording of their cold call, and the transcript would be analyzed using LLMs to determine the effectiveness of the pitch. The team’s goal was to expand this solution to call centers, where all calls would be analyzed to detect customer sentiment and provide representatives with tips on how to improve the customer experience.

Another startup, Hear Me Out, presented an AI-driven customer service pipeline that optimized “call-matching and visibility.” Their solution consisted of “micro-services” such as a call bot powered by LLMs and a call analytics service. The team emphasized the importance of determining the caller’s sentiment and using generative AI to generate an “agreeability score.” They also planned to use sentiment analysis to automate the creation of an overall call score.

What struck me about these demos was how familiar they sounded. The concepts being presented were not new – they are, in fact, the core underpinnings of Conversational AI and Conversational Intelligence platforms that have been around for years. It seemed like these young startup founders had just discovered how impactful analyzing customer interactions could be for a business. Thanks to the availability of LLMs–a technology many of us could only dream about when we were building CX solutions–this new generation of entrepreneurs are able to create products that address the need with remarkable ease.

Case in Point: Bland AI Raises $16M to Automate Calls

Bland AI is an example of a new CX startup riding the wave of GenAI to revolutionize the way businesses interact with their customers. The company, which recently emerged from stealth with a $16M Series A financing round, has developed a platform that automates phone calls for enterprises using hyper-realistic AI agents, all completely built using LLMs. Businesses can create custom AI agents that can understand human emotion, speak any language, and represent a brand like a top employee.

Bland AI’s platform is designed to streamline a wide range of phone-based tasks, from customer support to sales to internal operations. The company’s AI agents can be trained on a company’s own phone calling data, and can integrate with other systems to provide a seamless experience.

LLMs Makes Everything Easy

What’s truly remarkable about this new generation of CX startups is that none of them would be possible without the magic of large language models (LLMs) and natural language prompts. Gone are the days of complex training of traditional NLP systems, which required vast amounts of sample utterances and were still unreliable. Gone are the days of dialogue management, which was needed to manage multi-turn conversations and ensure that the system stayed on track. LLMs have eliminated the need for all of this, making it possible to create conversational AI systems that are not only more accurate, but also more flexible and easier to use.

Meanwhile, most enterprises are still using legacy systems that are tightly integrated with their existing infrastructure, and are struggling to keep up with the pace of innovation. The contrast between the old and the new is stark.

The Next Three Years are Crucial

Start-ups (and the VCs that fund them) help detect, define, and refine new features and functions for enterprise infrastructure. BlandAI, Spark, and Hear Me Out join dozens of Voice AI specialists, like Vapi, Gridspace, Speechly (acquired by Roblox), FlipCX (nee RedRoute), to name a few, helped define the whole Voice AI category. Popularization of LLMs is accelerating acceptance, while raising expectations that voicebots can be quickly developed and effectively implemented.

Large enterprises have traditionally relied on established players like Nice, Genesys, Avaya, and more recently, solutions from tech giants like Amazon and Zoom. These incumbents have a deep understanding of the complexities of enterprise customer service and a proven track record of delivering reliable, scalable solutions.Yet being an incumbent has not guaranteed continued success. The fate of Nuance’s technology within Microsoft’s pantheon of solutions is a case in point. The acquisition was  completed in early 2022. At the time, Nuance’s had what many considered to be the most advanced and reliable Conversational AI system, which had long been the industry benchmark. But before the year was out, OpenAI would unveil ChatGPT and everything changed.

Enterprise IT abhors a vacuum. As “end-of-life” heralds “end-of-support” look to a wide variety of solution providers to enter the market. VCs, voting with their dollars, anticipate serious investment in new solutions. New entrants must emphasize more than just speed-to-deploy, and simplicity. I’ll be watching to see which alternatives assure business continuity, reliability and safety. Some aspects of migration are difficult, and those are the ones that new entrants must recognize and mitigate.

(Picture generated using Remix)

Categories: Articles