Competitive comparison
ArcGlass vs. Decagon
D
ParentDecagon AI, Inc.
Founded2023
HQSan Francisco, CA (offices in New York and London)
Employees300+ (Mar 2026)
FundingSeries D · ~$481M raised; $250M Series D led by Coatue and Index Ventures (Jan 2026)
Valuation / ARR$4.5B valuation (Jan 2026; employee tender at the same valuation, Mar 2026); ARR not disclosed by the company
Websitedecagon.ai
PricingNot public; quote-based, per conversation (most customers, per Decagon) or per resolved conversation.
Notable customers
Avis Budget Group
Chime
Deutsche Telekom
Duolingo
Oura
Square
Company data compiled from public sources; figures are approximate and may have changed since publication.
Recent developments
- Sep 2026: launched Generative Analytics (natural-language charts and Action Trackers over agent conversations)
- Sep 2026: launched Campaign Composer for proactive outreach and Autopilot governance for self-improving agents; expanded into Brazil and Benelux
- Mar 2026: completed first employee tender offer at a $4.5B valuation, open to 300+ employees
- Jan 2026: raised $250M Series D at $4.5B, led by Coatue and Index Ventures; added 100+ enterprise customers in the prior fiscal year
TL;DR. Decagon builds AI "concierge" agents that resolve customer inquiries on chat, email and voice, defined in natural-language Agent Operating Procedures (AOPs). It raised $250M at a $4.5B valuation in January 2026. Its Watchtower product reviews support conversations, AI or human, for compliance, sentiment and escalations. ArcGlass does not talk to customers. It reads the meetings, calls, email and chat your people have with customers and tells leaders what is working, what needs attention, and what to change in the message. The overlap is conversation review, and it is narrow: Decagon reviews the support queue, ArcGlass reads the wider relationship for leadership.
Strategic positioning
| ArcGlass | Decagon | |
|---|---|---|
| Buyer | COO and CEO first; also CMO, VP Sales, Customer Success, multi-location operators | Head of Support or CX, VP Operations, customer experience leadership |
| Headline value | "Stay close to customers. Without sitting in every meeting." | "The AI concierge for every customer" |
| Center of gravity | Reading human conversations and briefing leadership | AI agents that resolve support conversations, plus QA and analytics on them |
| Who talks to the customer | Your people. ArcGlass reads afterward and never replies to a customer. | Decagon's agents, with handoff to human agents when needed |
| Primary data | Meetings, calls, email and chat between your team and customers | Support conversations on chat, email and voice, AI-handled and human-handled |
| Recent direction | Leadership views: messaging and positioning analysis, account health trends, a free on-device notetaker (Listen) | Generative Analytics, Campaign Composer, Autopilot for self-improving agents, international expansion |
Decagon resolves the customer's issue and measures how that went. ArcGlass reads the conversations your people have with customers and tells leaders what they add up to. Decagon works the support queue. ArcGlass works the leadership view.
Overlap surface
1. Resolving customer inquiries Decagon wins
This is Decagon's core product and ArcGlass does not compete here.
- ArcGlass: ArcGlass has no customer-facing agent, no support inbox and no ticketing. It does not answer customers or resolve issues.
- Decagon: Decagon agents resolve inquiries end to end, calling your systems to do things like process returns, recover accounts or handle disputes. Workflows are written as AOPs in natural language, and customers report resolution rates of 70% to 95% on Decagon's site.
2. Voice Decagon wins
- ArcGlass: ArcGlass reads calls recorded in Gong or Chorus and meetings on Zoom, Meet and Teams. Its Listen notetaker records on the desktop only. None of this is a phone agent.
- Decagon: Decagon Voice runs AI agents on the phone line, built with ElevenLabs voice technology, connected to existing call center infrastructure and driven by the same AOPs as chat and email. Voice 2.0 followed in 2025.
3. Which conversations get read Different conversations
- ArcGlass: ArcGlass reads the conversations between your people and customers across the business: meetings, notetaker output (Fireflies, Fathom, Granola), calls from Gong and Chorus, Gmail and Outlook email, and Slack, Teams, Discord and Google Chat messages.
- Decagon: Decagon reads support conversations on the channels it serves (chat, email, voice), including human-agent conversations for QA through Watchtower.
A support ticket about a failed payment is Decagon territory. The account review where the customer's COO says they are evaluating alternatives is ArcGlass territory.
4. Conversation review and QA Decagon wins inside support
The closest real overlap between the two products.
- ArcGlass: ArcGlass reviews each meeting for account signals (Jump on this, Tread carefully, In good shape), topics with sentiment and evidence quotes, conversation health, and rep moments. It is written for leaders across sales, customer success and the executive team, not as a support QA scorecard.
- Decagon: Watchtower reviews every support conversation, whether handled by AI or a human agent, against custom criteria. It flags compliance risks, negative sentiment, low CSAT, escalations and upsell signals, with dashboards and drilldowns by channel and time window.
If you need QA on every support conversation, Decagon has it. If you need to know what your customer conversations across the business mean for the company, that is ArcGlass.
5. Insight for the rest of the company Different scopes
- ArcGlass: Customer Questions collects what customers ask across conversations. Product Signals shows strong reactions and unmet needs per product area, linked to source calls. Messaging analysis shows what lands and what falls flat.
- Decagon: Generative Analytics (Sep 2026) lets CX and business teams ask questions of Decagon-handled conversations in plain language, build charts, and track outcomes with Action Trackers.
6. Escalations and human handoffs Complementary
- ArcGlass: When a customer is handed off and the conversation continues by email, in Slack or Teams, or in a meeting with an account team, ArcGlass reads that part. Agent transcripts could also be sent in as another conversation source through a custom integration or webhook.
- Decagon: Decagon hands conversations to human agents when an AOP calls for it, and Watchtower flags escalations and unresolved issues for review. Its view ends at the support interaction.
Decagon closes the ticket. ArcGlass tells leadership what the conversations that needed a person have in common.
7. Account health and the leadership view ArcGlass wins for B2B leadership
- ArcGlass: Companies shows each account as At risk, Watch or On track with a one-line reason, trended weekly. Home gives each leader an Act now / Watch / Working list and a week-over-week view of what changed across customers.
- Decagon: Decagon reports on support performance: resolution and deflection, CSAT, escalations and QA flags. It is not built to roll up the health of named business accounts across sales and success conversations.
8. Buying and rollout Different models
- ArcGlass: Every plan starts with a free 14-day trial. A source connects in about two minutes, there is no rollout project, and people do not change how they work. Listen is free.
- Decagon: Decagon sells quote-based contracts priced per conversation or per resolution. Agents are configured with AOPs, tested with simulations, and tuned with experiments and Autopilot.
Coverage areas only one side has
Only ArcGlass
- Reads meetings, calls, email and chat between your people and customers, with no new recorder to roll out
- Per-meeting analysis: bottom line, account signals, topics with sentiment and evidence quotes, actors, suggested follow-ups
- Account health (At risk, Watch, On track) with a reason, trended weekly
- Daily leadership briefing and a narrative of the period
- Messaging analysis and positioning (claims versus what customers raise unprompted)
- Rep moments and per-rep trends for sales and customer success teams
- Write-back to Salesforce and HubSpot, and follow-ups to Google Calendar and Outlook
- ArcGlass Listen: free desktop notetaker that records and transcribes on-device
Only Decagon
- Customer-facing AI agents that resolve inquiries on chat, email and voice
- Agent Operating Procedures (AOPs) for defining agent workflows in natural language
- Voice agents on the phone line, connected to existing call center infrastructure
- Watchtower QA across every support conversation, AI or human
- Generative Analytics and Action Trackers over agent conversations
- Campaign Composer for proactive, compliant outreach
- Simulations, experiments and Autopilot for testing and improving agents
- Per-conversation or per-resolution pricing
Takeaways
- These are not substitute products. Decagon answers customers and reviews the support queue. ArcGlass helps the leader who is not in the conversation understand it. Buying one does not remove the need for the other.
- Decagon is well ahead on support automation. Agents, AOPs, voice and support QA are the product, and ArcGlass does not intend to build any of it.
- Watchtower is the closest overlap, and it stays inside support. It reviews support conversations for support leaders. ArcGlass reads sales, customer success and account conversations for the COO, CEO and CMO.
- The meeting point is the handoff. When a support agent hands a business customer to a person, the rest of the relationship often plays out in email, chat and meetings. That is the part ArcGlass reads.
- If you're choosing between them: pick Decagon if our problem is "our support volume is too high for our team, and we want AI to resolve most inquiries on chat, email and phone, with QA on every conversation." Pick ArcGlass if our problem is "our leaders cannot see what is being said in customer meetings, calls, email and chat, which accounts need attention, or whether our message is landing." Decagon resolves and reviews support conversations. ArcGlass reads the human conversations across sales, success and account teams, including escalations that continue by email, chat or meetings, and briefs leadership on what they show.
How ArcGlass thinks about the overlap
We do not position ArcGlass as a Decagon alternative. Decagon is solving a real bottleneck inside support, and Watchtower gives support leaders review coverage ArcGlass does not try to match.
An earlier version of this page described ArcGlass deciding whether a ticket goes to Decagon and routing work to other teams' tools. That is not what ArcGlass does. ArcGlass reads the conversations your people have with customers and tells the COO, CEO and CMO what is working, what needs attention, and what to change in the message. If Decagon is answering your customers, ArcGlass is useful for the conversations that still need a person.
Sources
- Decagon's $250 million commitment to the AI concierge future (Decagon, Jan 27, 2026) (accessed 2026-09-27)
- AI Customer Support Startup Decagon Valued at $4.5 Billion (Bloomberg, Jan 28, 2026) (accessed 2026-09-27)
- Decagon completes first tender offer at $4.5B valuation (TechCrunch, Mar 4, 2026) (accessed 2026-09-27)
- Decagon revenue, valuation and funding (Sacra) (accessed 2026-09-27)
- Decagon (company) (Wikipedia) (accessed 2026-09-27)
- Decagon home page (accessed 2026-09-27)
- Watchtower: Always-on QA for every conversation (Decagon, Jun 25, 2025) (accessed 2026-09-27)
- Introducing Generative Analytics (Decagon, Sep 18, 2026) (accessed 2026-09-27)
- Announcing Decagon Voice (Decagon) (accessed 2026-09-27)
- Pricing the AI Agent Economy (Decagon) (accessed 2026-09-27)
- Decagon blog (Campaign Composer, Autopilot, Brazil and Benelux expansion, Sep 2026) (accessed 2026-09-27)
Questions about this comparison? Reach out at hello@arcglass.io — we're happy to dig into specifics for your stack.