Call centers are embracing automation. There is debate as to either that is goodbut it is happening — and quite possibly accelerating.
According to research firm TechSci Researchthe global market for AI contact centers could grow to nearly $3 billion in 2028, up from $2.4 billion in 2022. Meanwhile, a recent survey found that roughly half of contact centers plan to adopt some form of artificial intelligence next year.
The motivation is rather obvious: call centers are trying to cut costs while scaling their operations.
“Companies with heavy call center operations, trying to scale quickly without the limitations of human contact center agents, are particularly receptive to adopting effective AI voice agent solutions,” entrepreneur Evie Wang told TechCrunch. “This approach not only lowers their overall costs, but also reduces wait times.”
Wang is one of its co-founders Redo AI, which provides a platform that companies can use to create AI-powered “voice agents” that answer customer phone calls and perform basic tasks such as scheduling meetings. Retell agents are powered by a combination of large language models (LLMs) optimized for customer service use cases and a speech model that gives 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 deal with high call volumes, such as telehealth company Ro. They can build voice agents using the platform’s low-code tools, or they can upload a custom LLM (eg, an open model like Meta’s Llama 3) to further customize the experience.
“We are investing heavily in the voice chat experience, as we see it as the most critical aspect of the AI voice agent experience,” Wang said. “We don’t think of AI voice agents as simple toys that can be created with a few lines of prompts, but rather as tools that can deliver meaningful business value and replace complex workflows.”
Retell worked pretty well in my short tests, at least on the calling side.
I set up a call with a Retell bot using the demo form on the Retell website. The bot walked me through the process of scheduling a hypothetical dentist appointment, asking questions like preferred date and time, phone number, and so on.
I can’t say the bot’s synthetic voice was the best I’ve heard in terms of realism — certainly not on par with Eleven Labs or OpenAI’s text-to-speech API. (Update: Wang tells me that Retell uses a custom ElevenLabs voice, which may explain the lower quality.) Wang, in Retell’s defense, said the team has focused primarily on reducing latency and handling spikes , such as interruptions that may occur in a conversation.
The delay is low: In my test, the bot responded almost without hesitation to my answers and follow-up questions. And he stuck to his script. Try as I might, I couldn’t confuse it or make it behave the way it shouldn’t. (When I asked the bot about my dental records, it insisted I speak to the office manager.)
So are platforms like Retell the future of call centers?
It can. For basic tasks like appointment scheduling, automation makes a lot of sense, which is probably why both startups and big tech companies offer solutions that compete head-on with Retell’s. (See Parloa, PolyAI, Google Cloud Contact Center AI, etc.)
It’s low-hanging—and seemingly profitable—hanging fruit. Retell claims to have hundreds of customers, all of whom pay per minute of voice agent conversation. Retell has raised a total of $4.53 million in capital to date, thanks to backers including Y Combinator (where the company was incubated).
But the jury is out on more complex questions, particularly given the tendency of LLMs to fabricate facts and go off the rails even with safeguards.
As Retell’s ambitions grow, I’m curious to see how the company tackles the many established technical challenges in the space. Wang, at least, seems confident in Retell’s approach.
“With the advent of LLMs and recent breakthroughs in speech synthesis, conversational AI is getting good enough to create really compelling use cases,” Wang said. “For example, with sub-second latency and the ability to interrupt AI, we’ve seen users speak in fuller sentences and converse as they would with another person. We strive to make it easier for developers to build, test, deploy and monitor AI voice agents, ultimately helping them achieve production readiness.”
