Decagon

What is Decagon?

Decagon is a San Francisco AI company that builds autonomous customer-service agents for large enterprises, resolving conversations end to end across chat, voice, email and SMS rather than just answering questions. Founded in 2023 by Jesse Zhang and Ashwin Sreenivas, it raised a $250 million Series D in January 2026 at a $4.5 billion valuation, one of the best-funded companies in AI customer support. Decagon sells directly to enterprises such as Chime, Duolingo, Hertz, Rippling and Notion through negotiated contracts rather than as a self-serve product, with no published pricing.

What does Decagon do?

Decagon agents are configured through Agent Operating Procedures (AOPs) — workflows a support team writes in plain English, which the platform converts into the logic the AI follows, covering tasks like issuing refunds, changing subscriptions or pulling account data from connected systems rather than just replying with text. Its Duet AI Partner analyzes real conversations to spot gaps and automatically drafts or refines AOPs, while Simulations and Versioning/Experiments let a team test a change at scale and A/B it against real traffic before rolling it out fully. Watchtower and Trace View give 24/7 monitoring and step-by-step decision tracing, so a team can see which knowledge, workflow or model call produced a given answer. Agents run across chat, voice and email from one intelligence layer, and connect into ticketing platforms, CRMs, knowledge bases and CCaaS providers. Decagon publishes named case studies rather than only marketing claims: Duolingo reports an 80% deflection rate, ClassPass a 95% cost reduction, Chime a 70% chat-and-voice resolution rate, and Hunter Douglas $1 million in revenue attributed to fully AI-handled conversations.

Key features

• Agent Operating Procedures (AOPs): Support teams author agent behaviour in plain English; Decagon converts it into the workflow logic the AI actually follows.
• Duet AI Partner: Analyzes real conversations to find gaps, then auto-drafts or refines AOPs to close them.
• Watchtower + Trace View: 24/7 conversation monitoring plus step-by-step decision tracing back to the knowledge, workflow or model call behind an answer.
• Simulations & Experiments: Test a change against simulated conversations at scale, then A/B it against real traffic before a full rollout.
• Omnichannel from one layer: Chat, voice, email and SMS are handled by a single agent definition rather than separate point tools per channel.

What are Decagon's use cases?

• High-volume support deflection: A consumer brand like Duolingo or ClassPass routes the bulk of inbound chat/voice volume to Decagon, resolving most conversations without a human agent.
• Proactive outbound resolution: An agent reaches out ahead of a known issue (e.g. a delayed rental at Hertz) rather than waiting for the customer to contact support.
• Revenue-generating conversations: Hunter Douglas reports $1M in revenue attributed to conversations Decagon’s agent handled fully on its own, turning support into a sales channel.

What does Decagon integrate with?

• Ticketing platforms
• CRMs
• Knowledge bases
• CCaaS providers
• Channels — chat, voice, email, SMS

Why use Decagon?

• Backed by a $250M Series D (Jan 2026) at a $4.5B valuation from Coatue, Index Ventures, a16z and Accel among others, one of the best-capitalized vendors in AI customer support.
• Named, published case studies with concrete numbers, Duolingo’s 80% deflection rate and ClassPass’s 95% cost reduction are stated results, not vague claims.
• G2 reviewers rate it 4.9/5 across 18 reviews, consistently praising fast implementation and a responsive team.
• AOPs let non-technical support staff author and adjust agent behaviour in plain English instead of waiting on engineering for every change.

Pros & cons

Decagon pricing

• Enterprise (custom)Custom

See current pricing on decagon.ai ↗Compare Decagon alternatives →

Decagon review

Decagon is one of the best-capitalized names in AI customer support — a $250M Series D in January 2026 valued it at $4.5B, and its published case studies back that up with real numbers: Duolingo’s 80% deflection rate, ClassPass’s 95% cost reduction, Chime’s 70% chat-and-voice resolution rate. Plain-English AOPs let support teams (not just engineers) shape agent behaviour, and Watchtower/Trace View give real visibility into what the agent is doing. The honest catches: there is no public pricing — third-party analysis estimates $74K–$600K+/year depending on volume — resolution-based billing has drawn complaints about how a “resolution” is counted, and reviewers say meaningful deployments still need dedicated engineering to configure AOPs, often over 4–12 weeks. Pick Decagon if you are an enterprise that wants a well-funded, fast-improving agent and can run a proper implementation; skip it if you need transparent self-serve pricing or a public API.

One of the best-funded AI support agents on the market, with real published results, but budget for a multi-week implementation and an opaque, resolution-based bill.

Frequently asked questions