Decagon

Decagon is an AI calls tool. Enterprise AI agent platform for customer support: natural-language workflows, pre-built integrations, A/B testing, QA and analytics across chat, voice and email channels.

Decagon is an enterprise platform for building AI agents that handle customer support across chat, voice and email. The company positions the product as an "AI concierge" rather than a ticket deflector: the stated goal is to resolve each customer's issue with enough context that the exchange feels individual, and the headline claim is that agents are built, optimized and scaled inside one system instead of through a patchwork of vendors and configuration languages. The core building block is Agent Operating Procedures, or AOPs. Instead of writing code or wrestling with a proprietary scripting syntax, teams describe agent workflows in natural language and refine them as business rules change. Around that sit pre-built connectors to existing systems, live A/B testing, simulation-based testing and QA, always-on quality monitoring through Watchtower, an insights and reporting suite that surfaces voice-of-customer themes, and Suggestions, which feeds AI-generated knowledge back into the agent. Duet acts as an AI partner for building self-improving agents, and the platform detects personal agents visiting a company's properties and offers a dedicated channel for serving them with permissions the business controls. One intelligence layer powers all three channels, so a conversation that starts in chat can continue in email or on a voice call with the same memory of the customer. Voice is pitched at natural dialog and brand customization; chat at reliable execution of complex workflows; email at always-on resolution. Decagon publishes case studies with Chime, Duolingo, ClassPass, Hunter Douglas, Notion, Hertz, Duolingo, Perplexity and others, and organizes industry material around financial services, travel and hospitality, health and wellness, technology, retail, telecommunications and media. This is infrastructure for support organizations at scale, not a self-serve tool for a solo founder. Buyers should expect a demo-led sales process, a real integration and data preparation effort, and ongoing ownership of the AOP library. It competes with other enterprise support automation suites on how quickly non-engineers can change agent behavior, and the natural-language workflow layer is the main argument Decagon makes for itself.