The archive · AI & Models · Strategic decision · 2023–2026
Decagon bets AI agents can carry full customer support; three years to a $4.5B valuation
Two founders with prior exits bet buyers want AI agents that act, not just chat; Decagon hit break-even in 2024 and a $4.5B valuation by March 2026.
Decagon
What the business is
AI 'concierge' agents that autonomously resolve customer support inquiries across chat, email and voice for large companies.
Starting capital：$35M across seed and Series A; Accel led the A with a16z, A* and Elad Gil participating
How it started
Jesse Zhang (founded Lowkey, acquired by Niantic) and Ashwin Sreenivas (sold Helia to Scale AI) co-founded Decagon in 2023. Everyone advised them to avoid customer support as too crowded, but they went in anyway.
What happened
By June 2024 Decagon had raised $35M, reached break-even and signed Eventbrite, Bilt and Substack; backers included Box CEO Aaron Levie. ARR crossed eight figures by late 2024 and the company was valued at $1.5B in June 2025.
How it ended up
January 2026: $250M Series D from Coatue, Index, a16z, Definition, Forerunner and Ribbit. March 2026: $4.5B valuation and a first employee tender offer, with 300+ staff and 100+ large customers including Avis Budget Group and Oura.
Background
Decagon sells AI 'concierge' agents that resolve customer support across chat, email and voice for enterprises. Jesse Zhang and Ashwin Sreenivas — each with a prior exit, Zhang's gaming app Lowkey to Niantic and Sreenivas's Helia to Scale AI — founded it in 2023. The prevailing advice, Zhang recalled, was not to pursue customer support because the space was too crowded with rivals like Parloa, Retell AI and Cognigy.
The founders went anyway, betting that companies would pay for agents that behave like real employees rather than smarter chatbots. Decagon's fine-tunable bots ingest a business's knowledge bases and historical conversations, learn from feedback, and integrate with other apps to take actions — processing refunds, categorizing messages, drafting support articles — while an analytics dashboard reviews and tags conversations to surface trends.
The strategy paid off quickly. By June 2024 Decagon had raised $35M across seed and Series A (Accel led; a16z, A* and Elad Gil participated), reached break-even, and counted Eventbrite, Bilt and Substack among clients. ARR passed eight figures by late 2024, the valuation hit $1.5B in June 2025, and after a $250M Series D led by Coatue and Index in January 2026, it completed its first employee tender offer at a $4.5B valuation in March 2026 with 100+ large customers, including Avis Budget Group, 1-800-Flowers and Oura.
What has to be true
- Two prior exits gave the founders credibility with enterprise buyers burned by overpromising chatbot vendors.
- 'Human-like' was defined as taking action — refunds, routing, follow-ups — which is where support budgets actually go.
- Reaching break-even before chasing growth let them raise on results rather than promises.
- Customer focus over demo polish was the stated strategy from year one, a deliberate contrast to flashy AI demos.
What can be applied
A crowded category is a signal of real demand, not a warning; compete on depth of job done and customer discipline, not on flash.
Aftermath
After the $4.5B employee tender, Decagon competes with Sierra, Intercom and Parloa in a market Gartner sizes at roughly 17 million contact-center agents worldwide. It stopped disclosing revenue once ARR passed eight figures in late 2024; the 2026 tender was framed partly as an AI-talent retention tool, echoing moves by ElevenLabs and Linear. As of September 2026 the company continues to scale its enterprise business.
Sources
- Decagon completes first tender offer at $4.5B valuation
- Decagon claims its customer service bots are smarter than average
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card