One of the hot categories in the generative AI space is customer service, which is really not surprising when you concentrate on the technology’s potential to lower contact center costs while increasing scale. Critics say AI-powered generative customer service technology could lower wages, result in layoffs and ultimately provide a more error-prone experience for end users. Instead, proponents argue that generative AI will empower – not replace – staff while allowing them to focus on more meaningful tasks.
Jesse Zhang is in the supporter camp. Of course, he’s a bit biased. Along with Ashwin Sreenivas, Zhang is co-founder Decagona generative AI platform for automating various elements of customer service channels.
Zhang is well aware of how stiff the competition is in the AI-powered customer service market, which incorporates not only tech giants like Google and Amazon but also startups like Parloa, Retell AI and Cognigy (which recently raised 100 million dollars). By one estimate by 2032, the value of this sector might be $2.89 billionin comparison with $308.4 million in 2022.
However, Zhang believes that each Decagon’s engineering expertise and go-to-market approach give it an advantage. “When we started, the dominant advice we received was not to use the customer service space because it was too crowded,” Zhang told TechCrunch. “Ultimately, what worked for us was aggressively prioritizing what customers want and staying focused on what customers can get value from. This is the difference between a real business and a flashy AI demonstration.”
Both Zhang and Sreenivas have technical backgrounds and have worked in each startups and larger technology organizations. Zhang was a software engineer at Google before becoming a trader at Citadel, a market-making company, and founded Lowkey, a social gaming platform that was acquired by Pokémon GO maker Niantic in 2021. Sreenivas was an implementation strategy specialist at Palantir before co-founding computer vision startup Helia, which he sold to Unicorn Scale AI in 2020.
Decagon, which sells mainly to “high-growth” enterprises and startups, is developing customer service chatbots. Bots, powered by first- and third-party AI models, might be fine-tuned and can leverage company knowledge bases and historical customer conversations to realize a higher contextual understanding of issues.
“When we started building, we realized that ‘human-like bots’ required a lot because human agents are capable of complex reasoning, taking actions and analyzing conversations after the fact,” Zhang said. “It’s clear from conversations with customers that while everyone wants greater operational efficiency, it can’t come at the expense of customer experience – no one likes chatbots.”
So how do Decagon bots like traditional chatbots? Well, Zhang says they learn from previous conversations and feedback. Perhaps more importantly, they’ll integrate with other applications to take actions on behalf of the customer or agent, corresponding to processing a refund, categorizing an incoming message, or helping write a support article.
On the back end, corporations gain analytics and control over Decagon bots and their conversations.
“Human agents are able to analyze conversations to spot trends and find improvements,” Zhang said. “Our AI-powered analytics dashboard automatically reviews and tags customer conversations to identify topics, flag anomalies and suggest additions to the knowledge base to better respond to customer queries.”
Currently, generative AI has a popularity for being, well, not entirely perfect and, in some cases, ethically compromised. What would Zhang tell corporations that are concerned that Decagon’s bots will tell someone to eat glue or write article stuffed with plagiarismor that Decagon will train its internal models on their data?
Basically, he says, don’t fret. “Providing customers with the necessary guardrails and monitoring their AI agents was important,” he said. “We optimize our models for our clients, but we do it in such a way that no data might be by chance shared with one other client. For example, a model generating a response for customer A would never have access to data from customer B.
Decagon’s technology – while subject to the same limitations as any other AI-based generative application – has recently attracted customers from big-name brands like Eventbrite, Bilt and Substack, helping Decagon break even. High-profile investors have also joined the enterprise, including Box CEO Aaron Levie, Airtable CEO Howie Liu and Lattice CEO Jack Altman.
To date, Decagon has raised $35 million in seed and Series A rounds including Andreessen Horowitz, Accel (which led the Series A), A* and entrepreneur Elad Gil. Zhang says the funding is getting used to develop products and expand Decagon’s workforce in San Francisco.
“The key challenge is that customers equate AI agents with previous-generation chatbots that don’t actually get the job done,” Zhang said. “The customer service market is saturated with legacy chatbots that have eroded lost consumer trust. New solutions of this generation must cut through the noise of existing operators.”