Decagon
Decagon’s privacy policy lists a San Francisco, California mailing address.
What it is
Decagon is a conversational AI platform for building and improving customer experience agents across chat, voice, email, SMS, and WhatsApp. It brings those channels into one intelligence layer, where teams can define agent workflows in natural language using Agent Operating Procedures. Agents can take workflow actions and hand conversations to people; the platform also supports multiple languages and API access. For development and oversight, Decagon offers simulated conversations, unit testing, versioning, A/B experiments, observability, and ongoing conversation quality checks. Its Duet tool analyzes customer conversations to find workflow gaps and help generate, test, and refine procedures. Connectors, APIs, MCP, custom tools, and SIP trunking support links to data, actions, escalations, and voice systems. Listed safeguards include encryption, data-retention controls for AI providers, bad-actor detection, and a supervisor model that checks for hallucinations. The platform runs on web and API. Pricing is paid and available on request; Decagon is headquartered in San Francisco, California.
For and against
For
- Supports chat, voice, email, SMS, and WhatsApp
- Agents can perform actions and hand conversations to people
- Supports multilingual service and API access
- Includes simulated conversations, testing, and observability
- Lists encryption and AI safeguards
Against
- Paid product with pricing available on request
The verdict
Decagon combines customer-facing channels with tools for building, testing, and monitoring agents. Teams should request pricing to assess whether its paid offering fits their needs.
Who it suits
Decagon suits enterprises looking to deploy customer service agents across messaging, email, and voice channels. It may fit teams that need workflow actions, human handoff, multilingual support, or API connectivity.
On the record
- Access controls
- Decagon lists role-based access control, single sign-on with Okta and Microsoft Entra, and short-lived, scoped JWT API tokens.decagon.ai · read 10 Oct 2026
- Agent development
- Duet analyzes customer conversations to identify workflow gaps and generate, test, and refine AOPs.decagon.ai · read 10 Oct 2026
- Agent workflows
- Agent Operating Procedures let teams define agent behavior in natural language and adapt workflows as business needs change.decagon.ai · read 10 Oct 2026
- AI data handling
- The security page says Decagon enforces zero-data retention with AI providers, so conversation data is not stored or used for training.decagon.ai · read 9 Oct 2026
- AI safeguards
- Decagon describes bad-actor detection, a supervisor model that checks responses for hallucinations, and Watchtower conversation reviews against custom criteria.decagon.ai · read 10 Oct 2026
- Channels
- The platform unifies customer experiences across chat, voice, and email.decagon.ai · read 10 Oct 2026
- Company
- Decagon’s about page names Jesse Zhang and Ashwin Sreenivas as co-founders and describes its focus as concierge customer experiences.decagon.ai · read 9 Oct 2026
- Custom connections
- Decagon supports MCP connectivity, APIs, custom tool integrations, and SIP trunking for voice.decagon.ai · read 10 Oct 2026
- Custom connectivity
- The site says agents can retrieve data and trigger actions through self-serve APIs and custom tool integrations.decagon.ai · read 9 Oct 2026
- Enterprise resilience
- The security page lists multi-region infrastructure, model redundancy, platform uptime SLAs, autoscaling, and auto-failover.decagon.ai · read 9 Oct 2026
- Integrations
- Decagon lists connectors for Salesforce, Intercom, Zendesk, Confluence, Contentful, Kustomer, Amazon Connect, and RingCentral.decagon.ai · read 10 Oct 2026
- Intended customers
- Decagon describes its platform as enabling leading enterprises to deploy AI agents across customer channels.decagon.ai · read 10 Oct 2026
- Named integrations
- Examples include Salesforce, Intercom, Zendesk, Confluence, Contentful, Kustomer, Amazon Connect, and RingCentral.decagon.ai · read 9 Oct 2026
- Optimization
- The platform includes testing, observability, experimentation, and analytics for improving agents.decagon.ai · read 9 Oct 2026
- Pricing model
- A Decagon pricing article describes per-conversation and per-resolution pricing options; it does not give prices.webflow2.decagon.ai · read 9 Oct 2026
- Product
- Decagon is a conversational AI platform for building, optimizing, and scaling customer service agents.decagon.ai · read 10 Oct 2026
- Security
- The security page states that data is encrypted with AES-256 at rest and TLS 1.2+ in transit, and that Decagon enforces zero data retention with AI providers including OpenAI and Anthropic.decagon.ai · read 10 Oct 2026
- Support
- The site lists support contact [email protected].decagon.ai · read 10 Oct 2026
