GTM Hub
What is Decagon?
Decagon is an AI concierge platform that enables enterprises to build, optimize, and scale AI agents that resolve customer support across chat, voice, email, and SMS. Agents are built on a proprietary engine and can be customized using natural-language Agent Operating Procedures (AOPs). The platform maintains cross-channel memory and integrates with existing support stacks including CRMs, helpdesks, and knowledge bases.
Decagon news
Decagon reaches $4.5 billion valuation
San Francisco-based Decagon valued at $4.5bn, triple its valuation from six months prior. Platform has handled 80M+ customer conversations; 100+ enterprises signed up in 2025 alone. Customers include Chime, Duolingo, Hertz, and Deutsche Telekom.
Source: AI Magazine ↗Decagon launches Voice 3, Personal Agent Gateway, Agent Modules, and Duet Apprentice
Decagon announces four major releases: Voice 3 with Chord voice model and duplex architecture for natural conversations; Personal Agent Gateway for customer AI agents; Agent Modules for industry-specific extensions; Duet Apprentice for business learning. Also introduces Missions for long-horizon task coordination.
Source: Decagon Blog ↗
Key features
Agent Operating Procedures (AOPs)
Define agent workflows in natural language rather than complex code, enabling non-technical teams to architect and iterate on agent logic
Duet AI Partner
AI assistant that analyzes customer conversations to identify workflow gaps and automatically generate, test, and refine AOPs. Duet Apprentice variant learns your business and works alongside teams
Multi-Channel Support
Single agent logic deployed across voice, chat, email, and SMS with consistent customer experience
Voice 3 with Chord
Proprietary voice model from Decagon Labs built specifically for customer conversations, featuring duplex architecture that processes incoming audio while speaking for more natural interactions
Cross-Channel Memory
Maintains customer context across sessions and channels with data portability and granular control
Personal Agent Gateway
Identifies customer personal AI agents and provides dedicated channels with clear permissions and controls
Agent Modules
Industry-tailored foundations for extending agents across journeys beyond support
Watchtower
24/7 monitoring and QA visibility into customer conversations against custom criteria
Outbound Voice
Proactive outbound calling capabilities for sales, support, and customer updates
Seamless Integrations
Pre-built connectors for Salesforce, Zendesk, Intercom, Kustomer, Confluence, Amazon Connect, RingCentral, and MCP-based open connectivity
Analytics Suite
Comprehensive insights into conversation patterns, team performance, and business outcomes
Decagon pricing
Decagon uses a per-conversation pricing model with higher rates for complex queries and voice interactions. Contact sales for enterprise pricing.
Pros & cons
Pros
- + Natural language AOP system reduces configuration complexity compared to traditional SDKs
- + Duet AI assistant compresses days of iteration into minutes
- + Single unified platform across chat, voice, email, and SMS reduces vendor fragmentation
- + Proprietary Chord voice model delivers natural-sounding conversations with 70+ language support
- + Seamless integration with existing enterprise tech stacks
- + Cross-channel memory maintains context across sessions
- + Pre-built industry-specific modules accelerate deployment
Cons
- − Proprietary pricing model requires custom quotes; no public tiering available
- − Enterprise-focused platform may have higher minimum requirements for smaller organizations
- − Relies on third-party LLMs (OpenAI, Anthropic) alongside proprietary models
Who is Decagon best for?
- • Enterprise customer service teams seeking to automate support across multiple channels
- • Organizations with high conversation volumes looking to reduce contact center costs
- • Brands prioritizing personalized, concierge-level customer experiences
- • Companies with complex workflows that benefit from natural-language configuration
- • Businesses operating across multiple geographies requiring 70+ language support
