Decagon Review: Features, Pricing & Alternatives

Enterprise AI agents for chat, email and voice, configured with plain-language operating procedures.

PaidCustomer SupportReviewed October 2026

What is Decagon?

Decagon builds AI agents that handle customer conversations on chat, email and voice for large companies. Instead of building rigid flows, teams describe how the agent should behave in Agent Operating Procedures, or AOPs, written in natural language. The agent can then follow those procedures, call connected systems and resolve complex requests, with tools for testing, quality assurance and analytics around it.

Decagon was founded in 2023 in San Francisco by Jesse Zhang and Ashwin Sreenivas. It has grown quickly with enterprise customers such as Chime, Duolingo, Square, ClassPass, Delta and American Airlines listed on its website. In January 2026 it raised a Series D that Bloomberg reported valued the company at $4.5 billion. In June 2026 it launched Duet Autopilot, which proposes and tests improvements to agents for human approval.

What sets Decagon apart is its focus on giving CX teams control without heavy engineering. AOPs let non-developers change behavior, Experiments run live A/B tests, simulations test changes at scale and Watchtower monitors quality continuously. Pricing is custom, with a choice of paying per conversation or per resolution, which Decagon says most customers settle on per conversation for predictability.

Key features of Decagon

Agent Operating ProceduresWrite agent instructions in natural language that compile into reliable workflows.
Voice, chat and emailOne agent platform for phone calls, live chat and email support.
Duet AutopilotFinds improvement opportunities, tests changes against past conversations and proposes them for approval.
ExperimentsRun live A/B tests on agent behavior to see what improves outcomes.
Simulations and QATest changes at scale before release and monitor quality with Watchtower.
Analytics and knowledge suggestionsSpot trends in conversations and get suggestions for missing knowledge.

Who is Decagon best for?

Large consumer brands handling high volumes of chat, email and calls.
Fintech and travel companies that need agents to follow strict procedures.
CX teams who want to edit agent behavior without waiting on engineers.
Enterprises replacing older chatbots that want measurable resolution rates and A/B testing.

Popular ways to use Decagon

Account support: Handle balance, card and account questions with procedure-driven guardrails.
Booking changes: Let the agent change or cancel reservations through connected systems.
Voice automation: Answer routine calls with a branded voice agent.
Subscription management: Process plan changes, pauses and refunds by policy.
Continuous improvement: Use Duet Autopilot to suggest and validate agent updates.

Decagon pricing (October 2026)

Plan	Price	What you get
Per-conversation	Custom pricing	Fixed rate for every incoming conversation, with volume pricing
Per-resolution	Custom pricing	Higher fixed rate per fully resolved conversation, no charge for escalations

Decagon does not publish prices. Contracts are sales-led, and the rate depends on volume, channels and the pricing model you choose.

Decagon pros and cons

Pros

Natural language AOPs give CX teams control
Strong testing, A/B and QA tooling
Covers chat, email and voice
Flexible per-conversation or per-resolution pricing

Cons

No public pricing or self-serve trial
Aimed at enterprises, not small businesses
Requires integration work to unlock actions
Young company with a fast-changing product

How to get started with Decagon

Request a demo on decagon.ai and share your volumes and channels.
Pick high-volume contact reasons for the first agent.
Write AOPs for those workflows and connect required systems.
Run simulations on past conversations before launch.
Launch, then use Experiments and Autopilot suggestions to improve.

Tips to get better results from Decagon

Compare per-conversation and per-resolution costs using your real data.
Keep AOPs short and specific, one workflow at a time.
Require human review for Autopilot changes that touch policy.
Measure resolution and CSAT together, not just deflection.

Decagon vs alternatives

Decagon’s closest rival is Sierra, another enterprise AI agent startup, with Ada as a longer-running alternative. All three use custom pricing and suit large deployments.

Teams that want a faster start can try Fin, which publishes per-outcome prices, or add AI to existing help desks with Zendesk AI or Salesforce Agentforce. See all best customer support tools.

Our verdict: is Decagon worth it?

Decagon is a strong option for large companies that want highly customizable AI agents and the testing tools to run them safely. Smaller teams, or those wanting published prices, should look at self-serve tools first.

Decagon FAQs

How much does Decagon cost?

Decagon uses custom pricing. Customers choose per-conversation or per-resolution billing, and rates are agreed with sales.

What are AOPs?

Agent Operating Procedures are natural language instructions that define how Decagon’s agent handles a workflow.

What is Duet Autopilot?

Launched in June 2026, it finds ways to improve an agent, tests changes against historical conversations and presents them for human approval.

Does Decagon support voice?

Yes. Decagon offers voice agents alongside chat and email.

Is there a free trial?

No self-serve trial is offered; you request a demo.

Pricing and features change often. Details on this page were last reviewed in October 2026. Always confirm current plans on the official Decagon website.