Decagon

Decagon

Decagon handles support conversations and connects them to business workflows.

Oximy ResearchUpdated 19 September 2026

Overview

A+
Transparency
4 / 10
Trust
99 / 100

Trains on your data

Free
No
Paid
No
Opt-out
Not captured

Incidents

0

0

Assessed SEP 21, 2026 · Checked by hand SEP 21, 2026

Decagon asserts SOC 2, ISO 27001, GDPR, HIPAA and PCI DSS, and its DPA security annex prohibits training and confirms provider ZDR and deletion on request.

Trust breakdown

A+

99 / 100

Overall trust score

Data safetyComplianceIncidentsLegal
Data safety100
Compliance100
Incidents100
Legal93

Data and privacy

Trains on your data

No training on user data

Storage regions

USAEU

Sub-processors

Google Cloud PlatformOpenAIAmazon Web ServicesAzureTwilioAnthropic PBCCohere Inc.VercelElevenLabsCartesia

Data retention

Third-party model providers operate under Zero Data Retention (ZDR) agreements that contractually prohibit the retention, logging, or use of Customer Personal Data for model training, finetuning, or any purpose beyond real-time inference.

Deletion: Decagon will delete Customer Personal Data upon the Customer’s written request, within the timeframe specified in this DPA and in accordance with Applicable Privacy Laws.

Powered by

OpenAI GPT-3.5OpenAI GPT-4OpenAI GPT-4oOpenAI GPT-4 TurboOpenAI o1-miniProprietary fine-tuned open-source LLMs
Proprietary model

Decagon uses a multi-model AI stack. They leverage OpenAI models (GPT-3.5, 4, 4o, 4 Turbo, and o1-mini) for various tasks, including complex decision-making and query rewriting after fine-tuning. They also train and fine-tune their own proprietary open-source LLMs, specifically on enterprise customer support conversations, with approximately 80% of inference traffic running on these in-house models as of March 2026. They use Together AI's inference engine for production inference and Modal for training and serving their real-time voice AI models, including custom draft models for speculative decoding. Microsoft Azure is used to host both off-the-shelf and fine-tuned models.

Integrations and access

Integrations

API: YesSalesforceIntercomZendeskConfluenceContentfulKustomerAmazon ConnectRingCentralPostHogSnowflakeGoogle BigQueryAmazon RedshiftPostgreSQL

Available on

Web
iOS
Android
macOS
Windows
Linux
Extension
CLI

Pricing

Enterprise

Custom

Decagon does not publicly list its pricing. It offers two pricing models: per-conversation (a fixed rate for every incoming conversation, with flexible pricing for higher volumes) and per-resolution (a higher fixed rate for each fully resolved conversation, with no charge for escalations). Pricing is custom-quoted based on factors like ticket volume, channel mix, integration complexity, workflow depth, and SLA requirements. Third-party estimates for annual contracts range from approximately $100,000 to $580,000, with a median around $432,750. There is also a reported annual platform fee of $50,000 before any usage charges. There is no free trial or self-serve option; all accounts require a demo and custom quote.

Prices as listed on SEP 21, 2026; check the vendor's page.

Moat and openness

6/ 10
Moat 6 out of 10

Moat

  • Proprietary Model
  • Proprietary Data
  • Network Effects
  • Switching Costmoderate
  • Unique UX
  • Distributionmoderate

Decagon exhibits some defensibility through its proprietary fine-tuned LLMs, trained on enterprise customer support conversations, and its specialized Agent Operating Procedures (AOPs) for defining AI agent workflows. The extensive list of notable enterprise customers and integrations suggests a moderate switching cost once deeply embedded. However, the reliance on underlying OpenAI models and the competitive landscape in AI customer service prevent a stronger moat.

6/ 10
Openness 6 out of 10

Openness

  • Open model weights
  • Published research
  • Open source contributions
  • Transparency reports
  • Public safety evals

Decagon emphasizes transparency in its AI products, particularly for enterprise adoption. They aim to provide clear visibility into how their AI agents make decisions, referencing data sources, and drawing conclusions. Their tools, like 'Trace View', offer granular insights into agent behavior and reasoning. They also provide user-friendly feedback systems and proactive improvement suggestions. However, there is no indication of open-sourced models, published research, or public safety evaluations.

Timeline

  1. 2026-07Decagon reached $100M in annualized revenue.
  2. 2026-03Decagon completed its first employee tender offer at a $4.5 billion valuation.
  3. 2026-01Decagon was valued at $4.5 billion following a $250 million Series D funding round.
  4. 2025-09Decagon launched Voice 2.0, an upgrade to its enterprise-grade AI phone support platform, Decagon Voice, in partnership with ElevenLabs. This version reduced latency by 65% and added cross-channel memory, outbound calling, SMS integration, and finer brand controls.
  5. 2025-06Decagon raised a $131 million Series C funding round at a $1.5 billion valuation.
  6. 2024-10Decagon completed a $65 million Series B funding round, valuing the company at $650 million.
  7. 2024-05Decagon completed a $35 million Series A funding round.
  8. 2023-08Decagon was founded by Jesse Zhang and Ashwin Sreenivas.

Company

Decagon

Founded
2023
HQ
San Francisco, California, United States

Popularity

Not disclosed

monthly visits

Notable customers

Avis Budget GroupMercado LibreDeutsche TelekomNotionDuolingoRipplingBiltEventbriteSubstackOura HealthAffirmChimeHertzAway TravelClassPassSimplePracticeHunter Douglas1-800-FLOWERS.COMRitual CosmeticsCurologyFlashfood

Decagon operates in the rapidly growing AI-powered customer service automation market, providing conversational AI agents for enterprise customer support across various channels.

Value and ROI

What it should move: It shortens resolution time and handles more requests without staff intervention.

Best fit

It fits teams that need one configurable agent across several support channels.

The catch

Decagon needs support content and system access. Test policy exceptions, handoffs, and actions that affect customer accounts.

  • Duolingo: Uses Decagon to handle customer chat questions for the Duolingo English Test.[1]
  • Substack: Uses Decagon to automate support workflows for readers and writers.[2]

In our stacks for

References