Support work across a conversation and its systems
Decagon builds AI agents for enterprise customer service. Its product connects a customer's request with operations such as a refund or shipment change, and analyzes conversations to help support teams identify recurring issues and gaps in their knowledge base. The company's launch described a stack combining third-party and fine-tuned models for different tasks. [1]
Decagon emerged from stealth in June 2024 with existing customer deployments. The relevant contribution was the working support platform and its integrations; the funding announced alongside it was a means to expand that product. A chatbot response and a completed refund involve different parts of the same service workflow. [1]
Procedures that business teams can edit
Decagon calls its configuration approach Agent Operating Procedures, or AOPs. Business users describe how an agent should handle a situation in natural language, while engineers control the code and integrations that carry out the underlying actions. This divides ownership between people who know the service policy and people who maintain the systems it touches. [2]
AOPs are part of Decagon's product approach, not an independently governed agent standard. The company's account explains how teams configure and revise behavior; claims about faster deployment or better outcomes remain vendor claims until demonstrated for a particular implementation. [2]
Sources
- Decagon launches with $35m raised from Accel and a16z
Decagon · Jun 18, 2024
- Agent Operating Procedures: Decagon’s approach to AI agents for CX
Decagon · Apr 8, 2025