Decagon, a startup that develops AI agents, has raised $65 million. Its clients are cutting support staff by 80%
Decagon CEO Jesse Zhang claims that their AI service was compared with solutions from Salesforce, and Decagon came out the winner.
Disclaimer: this is a free translation notes Forbes publications. The translation was prepared by the editors of Technocracy. In order not to miss the announcement of new materials, subscribe to “Voice of Technocracy” — we regularly talk about news about AI, LLM and RAG, and also share useful tutorials and current events.
You can discuss the pilot or ask a question about LLM Here.
Jesse Zhang didn't set out to create a product that he thought people would like. Re-creating a startup, he wanted to develop software that companies could quickly implement and get real value from. In August 2023, Zhang teamed up with Ashwin Srinivas, whom he met at a founders' meeting in Utah, and interviewed dozens of startups to create a product that was truly in demand and that companies would be willing to pay for – an AI system that could solve routine customer service tasks. . Decagon was born soon after.
Just over a year later, companies like Notion, Duolingo, Rippling, Bilt, and Substack are already using Decagon to answer questions about their apps, process returns, cancel subscriptions, resolve disputes, and replace credit cards.
However, getting those clients wasn't easy, Zhang tells Forbes. Nearly all of Decagon's customers ran what they called “validation tests,” comparing its software to rivals like Salesforce to determine which chatbot solved problems better. In each of these tests, Decagon's AI chatbot consistently emerged victorious, Zhang noted. Such assessments are becoming more common as companies test different AI models. For example, Intercom recently switched from OpenAI's GPT to Anthropic's Claude after a series of tests of its own, Fortune reported.
Decagon today announced it has raised $65 million in a Series B round led by Bain Capital Ventures with participation from investor Elad Gil and leading venture capital firms Accel and Bond Capital. With $100 million in total funding, the startup is now valued at $650 million. The company plans to use the funds to expand its team, attract new customers and add voice functionality to the product.
The fresh investment comes at a time when AI agents—software that can not only answer queries but also perform specific tasks—are becoming increasingly popular. OpenAI CEO Sam Altman is betting that AI agents will become part of everyday life within the next year. Recent technical advances in AI, such as the ability to multitask, are leading to startups developing AI agents that can write code or perform complex legal research.
“When you combine high logical thinking ability with the use of other software tools such as web browsers, we are on the verge of creating agents that can do much of the mundane work that people typically want to do away with,” said Aref Hilali, partner at Bain Capital. Ventures, in an interview with Forbes.
Companies using Decagon chatbots are already adjusting their hiring plans. Some, like card and loyalty program provider Bilt, have reduced their support teams from hundreds of contractors to 65 people, saving millions of dollars. Another client, which Zhang did not name, cut its support team by 80%.
Decagon's AI agents are based on large language models from companies such as OpenAI, Anthropic and Cohere, and are trained on enterprise data: instructions, blogs and past customer conversations. A major challenge in training is that information often becomes outdated, so Decagon has developed a system to keep data up to date. As a result, the chatbot acts almost like a good human agent, Zhang noted.
Zhang, 27, argues that the success of their software lies not only in providing reliable information, but also in the ability for companies to control and customize AI responses to customer queries.
“I think they have wisely created a transparent management system so that their AI agents are not black boxes,” said investor Elad Gil in an interview with Forbes. “The customer can see what's going on, understand it, and change it so the system does exactly what it needs to do.”
Automating customer service is not a new idea. This is one of the first applications to integrate ChatGPT. This is also one of the most problematic areas for consumers, who often want to switch to human interaction after a few minutes of interacting with AI. Some companies, such as Discover, offer customers the ability to contact a person at any time.
However, companies like Decagon believe that AI agents, unlike old chatbots with programmed responses, can solve problems without human intervention.
“The frustration that people were experiencing was because the systems were not working properly. Now they are working,” said Hilali of Bain Capital Ventures.
