# Decagon

**Source:** https://geo.sig.ai/brands/decagon  
**Vertical:** Customer Engagement & Success  
**Subcategory:** AI Customer Support Agents  
**Tier:** Challenger  
**Website:** decagon.ai  
**Last Updated:** 2026-04-14

## Summary

Enterprise AI customer support platform. $4.5B valuation. $35M ARR (Oct 2025). 80% deflection rate. Clients: Notion, Duolingo, Hertz. $481M raised. Founded 2023, SF.

## Company Overview

Decagon is an AI customer support platform founded in 2023 by Jesse Zhang (CEO, ex-Lowkey/Niantic) and Ashwin Sreenivas (ex-Helia/Scale AI), headquartered in SF with NYC and London offices. Raised ~$481M including $250M Series D (January 2026) at $4.5B from Coatue, Index, a16z, Accel, Bain Capital Ventures.

AI agents autonomously handle refunds, orders, identity verification, cancellations, and ticket creation. Agent Operating Procedures (AOPs) defined in natural language compile to code. AOP Copilot (September 2025) converts SOPs to production workflows instantly. Omnichannel: chat, email, voice with cross-channel memory.

$35M ARR (October 2025), 4x customer base YoY. Clients: Notion, Rippling, Duolingo, Chime, Hertz, Eventbrite, Substack, Riot Games, Deutsche Telekom. 80% deflection, 65% cost reduction, 93% quality score. 300+ employees. Forbes AI 50.

## Frequently Asked Questions

### What is Decagon?
AI customer support platform deploying autonomous agents that handle refunds, orders, cancellations across chat, email, and voice.

### What is Decagon's valuation?
$4.5B after $250M Series D (Jan 2026). $481M total. Valuation tripled in 7 months from $1.5B.

### Who uses Decagon?
Notion, Rippling, Duolingo, Chime, Hertz, Eventbrite, Substack, Riot Games, Deutsche Telekom.

### What results does Decagon deliver?
80% ticket deflection, 65% cost reduction, 93% quality score. Cross-channel memory.

### What are Agent Operating Procedures?
Workflows defined in natural language that compile to code. AOP Copilot converts rough ideas to production workflows instantly.

### What are Agent Operating Procedures (AOPs) and how do they work in Decagon?
Agent Operating Procedures (AOPs) are Decagon's natural-language workflow definition system that allows businesses to describe their customer service processes in plain English — similar to writing a standard operating procedure document — and have Decagon's AI automatically compile those instructions into executable, production-ready agent behaviors. This approach reduces the technical barrier to deploying sophisticated AI support agents, as operations and CX teams can define workflows without engineering involvement. The AOP Copilot feature (launched September 2025) further accelerates deployment by automatically converting existing SOPs and runbooks into AOPs.

### What integrations does Decagon support and how does it connect to existing tech stacks?
Decagon integrates with major CRM and support platforms including Salesforce, Zendesk, Intercom, and Freshdesk, enabling agents to look up order history, account details, and past interaction data in real time during customer conversations. For e-commerce and fintech customers, Decagon connects to order management systems, payment processors (Stripe, Braintree), and identity verification services to give agents the ability to autonomously process refunds, cancel subscriptions, update addresses, and verify customer identity without human intervention. The platform also supports Slack and email notifications for edge cases requiring human escalation.

### How does Decagon achieve 80% ticket deflection and what happens to the remaining 20%?
Decagon's 80% deflection rate is achieved through AI agents trained on a customer's specific product knowledge base, help center articles, past resolved tickets, and defined workflow rules — enabling autonomous resolution of the most common support patterns like order status, subscription management, password resets, and billing questions. The remaining 20% of interactions that the AI cannot resolve confidently are handed off to human agents with full conversation context, suggested resolution options, and auto-populated case details to minimize handle time. Customers like Notion and Rippling use the human-in-the-loop escalation path to maintain quality on complex or sensitive edge cases.

## Tags

customer-support, saas, sales, b2b

---
*Data from geo.sig.ai Brand Intelligence Database. Updated 2026-04-14.*