Automation is becoming increasingly prevalent in call centers, and while opinions vary on its benefits, the trend is clear and likely to gain momentum.
TechSci Research, a market research firm, projects that the global market for AI in contact centers could increase to almost $3 billion by 2028, up from $2.4 billion in 2022. Additionally, recent data indicates that about 50% of contact centers are considering implementing AI technology within the coming year.
The rationale is straightforward: Call centers aim to minimize expenses while expanding their capabilities.
“Businesses that rely heavily on call center functions and are eager to expand rapidly without being limited by human agents are very open to integrating AI voice agent technologies,” said Evie Wang, an entrepreneur, in an interview with TechCrunch. “Not only does this strategy help in cutting costs, but it also helps in reducing the waiting time for customers.”
Wang, a co-founder of Retell AI, has developed a platform that enables companies to deploy AI-driven “voice agents” to handle customer calls and execute simple tasks like booking appointments. These agents utilize sophisticated language models fine-tuned for customer service scenarios, coupled with a speech model that vocalizes the text produced by these language models.
Retell AI serves various clients, including contact center operators and small to medium-sized enterprises that experience high call volumes, such as the telehealth firm Ro. Clients have the option to construct voice agents using Retell’s user-friendly tools or to import a bespoke language model (for example, Meta’s open-source Llama 3 model) to customize the service further.
Wang emphasizes the importance of the voice interaction quality, stating, “We place significant investment in the voice conversation aspect, as it’s the most vital element of the AI voice agent’s functionality. We perceive these AI voice agents not as simple playthings created with minimal input but as robust instruments that deliver considerable benefits to companies and can substitute intricate processes.”
During my brief evaluation, Retell proved to be effective, particularly in terms of interacting with calls.
I initiated a call with a Retell bot through the company’s demo request form. The bot guided me through setting up a mock dentist appointment, inquiring about details such as my preferred date and time, contact number, and more.
While the bot’s artificial voice didn’t impress me as the most lifelike I’ve encountered — it didn’t match the quality of Eleven Labs or OpenAI’s text-to-speech technologies — Wang explained that Retell has prioritized minimizing delays and managing unexpected scenarios, such as interruptions during a call.
The response time was impressively quick: the bot answered my queries and follow-up questions almost instantaneously during my trial. It adhered strictly to its programmed dialogue. Despite my attempts, I was unable to disrupt its operation or cause it to act inappropriately. (For instance, when I inquired about my dental history, the bot directed me to consult with the office manager.)
Could platforms like Retell represent the future of call centers?
Possibly. Automation seems particularly suited for straightforward tasks such as scheduling appointments, which is likely why numerous startups and major tech companies are developing competing technologies similar to Retell’s offerings (like Parloa, PolyAI, and Google Cloud’s Contact Center AI).
This area appears to be an accessible and potentially profitable market segment. Retell boasts a customer base in the hundreds, all of whom pay by the minute for voice agent interactions. To date, Retell has secured $4.53 million in funding, with contributions from investors including Y Combinator, where the company was nurtured.
However, the effectiveness of handling more complex inquiries remains uncertain, especially considering the propensity of large language models to fabricate information or deviate from expected responses, even with protective measures.
As Retell aims higher, it will be interesting to watch how the firm tackles the numerous established technical hurdles in this field. Wang, for her part, appears optimistic about Retell’s strategy.
“Wang stated, “The emergence of large language models and advancements in voice synthesis have brought conversational AI to a level where it can generate truly compelling applications. For instance, with nearly instantaneous response times and the capability to interrupt the AI, we’ve noticed that users are engaging in more extensive dialogues, interacting as they would with a human. Our goal is to simplify the process for developers to create, evaluate, implement, and oversee AI voice agents, ultimately aiding them in reaching a state of operational readiness.”