Call centers are increasingly automated. There's debate about whether that's a good thing, but it's happening and probably accelerating.
The global market for contact center AI could grow from $2.4 billion in 2022 to nearly $3 billion in 2028, according to research firm TechSci Research. Meanwhile, a recent survey found that nearly half of contact centers plan to implement some form of AI. next year.
The motive is clear. That's because call centers are trying to cut costs while expanding their operations.
“Companies with high-volume call center operations that want to scale quickly without the constraints of human contact center agents are excited to adopt an effective AI voice agent solution. ” entrepreneur Evie Wang told TechCrunch. “This approach not only reduces overall costs but also reduces wait times.”
Wang is one of the co-founders of Retell AI, a company that helps businesses create AI-powered “voice agents” that answer customer calls and perform basic tasks such as scheduling appointments. We provide a platform that you can use. Retell's agents are powered by a combination of large-scale language models (LLMs) fine-tuned for customer service use cases and speech models that give voice to the text generated by the LLMs.
Retell's customers include some contact center operators, as well as small and medium-sized businesses that regularly handle a high volume of calls, like telemedicine company Ro. You can use the platform's low-code tools to build voice agents or upload custom LLMs (e.g., open models such as Meta's Llama 3) to further tailor the experience.
“We are investing heavily in the voice conversation experience because we believe it is the most important aspect of the AI voice agent experience,” said Wang. “We don't think of AI voice agents as just toys that can be created with a few lines of prompts. Rather, we think of them as tools that can provide significant value to businesses and replace complex workflows. ”
Retell worked well enough in my brief testing, at least on the calling side.
I arranged a call with the Retell bot using the demo form on Retell's website. The bot guided you through the process of scheduling a virtual dentist appointment, asking questions such as your preferred date and time, and your phone number.
I can't say that the bot's synthesized voice was the best I've heard in terms of realism. It certainly wasn't on par with Eleven Labs or OpenAI's text-to-speech API. Wang defended Retell, saying the team has primarily focused on reducing wait times and addressing special cases such as interruptions that can occur during conversations.
Low latency: In my testing, the bot responded to my answers and follow-up questions with little hesitation. And it was true to the script. Try as I might, I couldn't confuse it or make it behave in a way it shouldn't. (When I asked the bot about my dental records, he requested to speak to the office manager.)
So are platforms like Retell the future of call centers?
perhaps. For basic tasks like scheduling appointments, automation makes a lot of sense. This is probably why both startups and big tech companies are offering solutions that directly compete with his Retell solution. (See Parloa, PolyAI, Google Cloud's Contact Center AI, etc.)
It's a low-hanging fruit and seemingly profitable. Retell has hundreds of customers, all of whom claim to pay per minute of voice agent conversation. Retell has raised a total of $4.53 million in funding to date, courtesy of backers including Y Combinator (where the company was founded).
But jurors have balked at more complex questions, especially given LLM's tendency to fabricate facts and go erratic even when safeguards are in place.
As Retell's ambitions grow, I'm interested to see how the company navigates the many established technological challenges in the space. At least Wang seems confident in Letel's approach.
“With the advent of LLM and recent advances in speech synthesis, conversational AI is powerful enough to create some very exciting use cases,” Wang said. “For example, we observed that with sub-second delays and the ability to interrupt the AI, users spoke in fuller sentences and conversed like other people. We strive to make it easy to build, test, deploy, monitor voice agents, and ultimately make them production-ready.”