How This Entrepreneur Built A $1.5 Billion AI Unicorn In One Year

In the frenzied world of AI startups, where hype often outpaces substance, Jesse Zhang’s story stands out like a glitch in the matrix—raw, improbable, and blisteringly fast.

At just 28 years old, the Harvard dropout has built Decagon, a San Francisco-based AI powerhouse that’s reimagining customer service as a seamless, empathetic conversation between humans and hyper-intelligent agents.

What started as a bold hunch in 2023 has exploded into a $1.5 billion unicorn, serving over 100 enterprise clients like Notion, Duolingo, Hertz, and Eventbrite.

In under two years, Decagon rocketed from stealth to eight-figure annual recurring revenue (ARR), raising $231 million along the way. But this isn’t a tale of overnight luck; it’s a masterclass in grit, customer obsession, and weaponizing speed in an industry where giants like Salesforce and Sierra loom large. Buckle up—here’s how Zhang pulled it off.

The Prodigy from Boulder: Roots in Math and Mayhem

Jesse Zhang wasn’t born with a silver algorithm in his mouth, but his trajectory screams “destined for disruption.” Raised in Boulder, Colorado, young Jesse was the kid dominating math decathlons and competitive programming contests, the kind of nerdy pursuits that foreshadow a life hacking systems.

“I had a pretty mathy background,” he later quipped in a podcast, a nod to the analytical edge that would define his career. By high school, he was already building software side hustles, blending code with an unquenchable curiosity about how people interact with tech.

Harvard came calling for computer science, but Zhang treated it like a speedrun. He graduated a year early in 2018, convinced the ivory tower’s curriculum was “not very relevant to startup building.” No time to waste: At 20, he dove headfirst into entrepreneurship, co-founding Lowkey, a consumer app for gamers to share highlight clips.

It was a chaotic baptism by fire. “I spent six years struggling to build that consumer business,” Zhang reflected, learning the hard way about user feedback loops and the perils of ignoring market signals. Lowkey sold to Niantic (the Pokémon GO folks) in 2022, a modest exit that netted experience but no fireworks. Zhang emerged wiser, scarred, and laser-focused: Next time, he’d build for enterprises, where pain points are deep-pocketed and predictable.

Enter Ashwin Sreenivas, his co-founder and CTO—a fellow serial entrepreneur with a prior exit under his belt. The duo bonded over a shared disdain for bloated customer service ops. “We just felt like there was no time to waste,” Zhang said of their early brainstorming sessions. In late 2022, amid the generative AI boom sparked by ChatGPT, they zeroed in on a $12 billion market ripe for revolution: customer support.

Call centers were bleeding—40% annual churn, endless escalations, and frustrated users stuck in IVR hell. Why not deploy AI agents that could handle refunds, verifications, and chit-chat with human-like finesse?

Stealth Mode: From LinkedIn Cold Calls to Product-Market Fit

Decagon was born in stealth in early 2023, a scrappy operation with Zhang and Sreenivas pounding the pavement. No fancy office—just relentless customer discovery. They fired off hundreds of LinkedIn messages, interviewing prospects over coffee and Zoom.

“What AI product would you actually pay for?” was their North Star question. The verdict was unanimous: Enterprises craved agents that could deflect 70-80% of tickets, slashing costs by up to 95% while boosting satisfaction 3x.

But talk is cheap. Zhang’s Lowkey scars taught him to “optimize for execution speed early.” For the first 18 months, they ignored long-term visions and distractions, hyper-focusing on deployments. They leveraged off-the-shelf LLMs from OpenAI, Anthropic, and Cohere as a foundation, then layered on proprietary magic: Agent Operating Procedures (AOPs).

This breakthrough let non-technical ops teams tweak agent behavior in natural language—like scripting empathetic responses for refunds—while devs controlled the code. Real-time dashboards tracked performance across channels (chat, email, voice), turning guesswork into data-driven iteration.

The name “Decagon”? A geeky flex—10 sides for their “10x better” mantra, born from Zhang’s math roots and a brainstorm with Sreenivas. By mid-2023, they emerged from stealth with seed funding from Accel ($16.6M), betting on conversation-based pricing: Charge per interaction, not per seat, for transparency and scalability.

Early wins snowballed. Notion signed on for seamless support; Duolingo’s users got instant streak fixes. Decagon’s agents weren’t just bots—they were “concierge-level,” handling escalations with nuance. Revenue hit seven figures in six months, then eight figures in ARR by year one. “The ROI is really easy to justify internally,” Zhang noted. “If you’re able to take [deflection] to 70-80%, that’s huge.”

The Unicorn Sprint: Funding Frenzy and Head-Shaving Bets

Silicon Valley noticed. Series A ($30M) and B ($65M) flew in from Bain Capital Ventures and Elad Gil, valuing Decagon at $650M by late 2024. But the real fireworks hit in June 2025: A $131M Series C co-led by Andreessen Horowitz and Accel, catapulting valuation to $1.5B—oversubscribed 5x. Total haul: $231M. Forbes pegged 2024 ARR at $10M, surging past $30M in 2025 (with Q3 tripling YoY). Zhang and Sreenivas each hold ~25% stakes, netting ~$370M paper fortunes.

Culture? Unhinged and unbreakable. Accel partner Ivan Zhou shaved his head in the office after a 10x revenue bet paid off—Zhang and Sreenivas wielded the clippers. Hitting milestones now means Hawaii trips for the 200-person team. “It’s going to be expensive,” Zhang joked. Hiring? “Intelligence over direct experience,” he preaches—raw smarts trumps rote resumes.

Decagon’s edge sharpened in “agent battles”—bake-offs where prospects pit their AI against rivals. Decagon won every time, from Hertz’s rental refunds to Oura’s sleep data queries. Voice AI? They hack limitations by transcribing calls to text, processing with LLMs, then voicing back—smooth as silk.

The Shadow of Giants: Can Decagon Outrun the Pack?

Yet, Zhang knows the arena’s brutal. Salesforce’s Agentforce rakes $440M quarterly; Intercom boasts 7,000 customers and 5x Decagon’s revenue; Zendesk eyes $200M from 20,000 AI users.

Sierra, backed by Bret Taylor, values at $10B. “The market is more competitive than ever,” admits Zendesk’s CTO. Zhang’s riposte? Agility. “Compared to the big Sierras and Salesforces, we have a bit of a younger team,” he admits, but that fuels 10x faster shipping. They’re already raising again, eyeing $3B+ valuation.

Zhang’s philosophy? “Success comes from matching execution with the right market.” Ditch playbooks that don’t fit; chase commercial wins over tech perfection. In a No Priors podcast, he urged founders: “Don’t overthink—go sell and get paid, or pivot.” For customer trust in AI? “It’s hard to earn, harder to win back”—so build empathetically, iterate obsessively.

The Horizon: AI’s Concierge Future

Today, Decagon powers tens of millions of interactions, deflecting tickets and delighting users. Zhang envisions “every brand with a dedicated AI agent that knows customers deeply.” But he’s no dreamer—he’s the kid who shaved an investor’s head and shipped a unicorn in a year. As AI commoditizes models, apps like Decagon win by owning the human touch.

Jesse Zhang’s odyssey proves it: In AI’s gold rush, the fortune favors the focused. From Boulder math whiz to Forbes 30 Under 30 AI honoree, he’s not just building software—he’s rewriting the rules of connection. And at this velocity, watch out: The next $4B round might just be his victory lap.