Decagon did not wake up one morning and decide to raise a $250 million Series D at a $4.5 billion valuation. This was a slow, deliberate build that started in August 2023 when Jesse Zhang and Ashwin Sreenivas met at an Andreessen Horowitz retreat in Utah, two founders with exits behind them and zero patience for small ideas. They kept hearing the same thing from operators across travel, fintech, healthcare, retail, and telecom. Customer support was breaking under its own weight, and nobody trusted AI to touch it without guardrails. Decagon was born inside that tension, not outside it.
Fast forward to January 28, 2026. Decagon announces a $250 million Series D led by Coatue Management and Index Ventures, with Lucas Swisher and Sofia Dolfe backing the conviction. Andreessen Horowitz, Accel Partners, Bain Capital Ventures, Forerunner Ventures, Ribbit Capital Management, T.Capital, and a deep bench of returning believers followed. The round pushes total funding to $481 million and cements a valuation that tripled in under six months, moving from $1.5 billion to $4.5 billion without the theatrics.
The product story is where the name earns its geometry. Decagon builds AI agents that do not freestyle. They operate inside Agent Operating Procedures, written in natural language, audited in real time, and controlled by the enterprise. These agents run across chat, email, SMS, and voice, with voice launching in February 2025 through a partnership with ElevenLabs. The result is not novelty. It is performance. Average deflection rates above 80 percent. One customer at 91 percent. ClassPass cutting cost per conversation by 95 percent. Chime reducing contact center costs by more than 60 percent while doubling Net Promoter Score.
Customers read like a global operations map. Avis Budget Group, Hertz, Chime, Affirm, Mercado Libre, Deutsche Telekom, Oura Health, Duolingo, Notion, Figma, Rippling, Webflow, Grubhub, Eventbrite, Substack, and more than 100 enterprise customers added in 2025 alone. ARR crossed $30 million in 2025 and kept climbing. Headquarters sit on Market Street in San Francisco, with teams in New York and London, quietly scaling while legacy platforms debate roadmaps.
This round is not about survival or speculation. It is about timing. Legacy customer support software was built for tickets. Decagon is built for conversations, memory, and accountability. Model agnostic, deeply integrated, and enterprise grade by design. Jesse Zhang brings the operator math of Citadel and Google. Ashwin Sreenivas brings the systems discipline of Palantir and Scale AI. The board reads like a syllabus in category creation, not hype management.
A decagon has ten sides. Most companies barely manage one. Watching this one stack precision, capital, and execution raises a sharper question for anyone still selling “AI powered” support. When the concierge shows up and actually solves the problem, how long does the lobby stay crowded.
