Decagon is an American company specializing in artificial intelligence agents for customer service. Available at decagon.ai, its platform is aimed at organizations looking to handle some of their support requests automatically while retaining human teams for complex situations. Its positioning goes beyond that of a document-based chatbot: the company develops agents capable of finding information, following a client’s specific procedures and triggering certain operations in its systems.
A company born amid the rise of generative AI
Decagon was founded in 2023 by Jesse Zhang and Ashwin Sreenivas in San Francisco. The two entrepreneurs launched the company at a time when large language models were opening up new possibilities for automating interactions with users. They chose to focus on customer service, a function that brings together high volumes of conversations, extensive documentation and repetitive tasks.
The company has attracted investors such as Accel and Andreessen Horowitz. It lists Notion, Eventbrite and Bilt among its clients. These references illustrate its foothold among digital businesses whose users expect support integrated directly into their online experience. The commercial challenge is to move from convincing demonstrations to reliable operation in production environments.
Agents connected to knowledge and business tools
The platform combines generative AI models with a company’s information sources: help centers, internal documentation and data accessible through integrations. This connection is intended to enable agents to provide contextualized responses rather than limit themselves to generic dialogue. Depending on the access granted and the procedures configured, they can also carry out tasks related to a customer request.
Decagon emphasizes an approach based on operational procedures that describe the agents’ expected behavior. Teams can thus guide the steps toward a resolution, define the circumstances requiring human intervention and analyze conversations. The goal is to make automation manageable by support leaders without requiring them to build the entire technical infrastructure themselves.
However, the value of the system depends on the quality of the available knowledge, the relevance of the integrations and permission controls. A plausible response is not enough when a request concerns an account, billing or a sensitive operation.
What’s next?
Decagon operates in a market where AI agent specialists and established customer service software providers converge. Its trajectory will depend on its ability to demonstrate lasting gains without compromising the quality of interactions. For its clients, the decisive criteria remain concrete: effective resolution of requests, traceability of actions, data protection and seamless handover to a human representative. The company’s development will depend as much on this operational control as on advances in language models.