Jesse Zhang

AI · Customer service Agent
United StatesAge ~29

Co-founder and CEO of Decagon · Founder of Lowkey

He built a gaming community in his first startup and sold it to Niantic; in his second, he skipped consumer hype for the AI customer service problem that is hardest to sell but closest to enterprise budgets.

Started while in collegeSerial founderScaled company$10M–$100M · 2024VC-backedSilicon ValleyFrom United StatesBorn 1997Top-school trainedForbes 30 Under 30 AI (Class of 2026)

Key numbers

Each figure includes its definition, date, and evidence level.

1997Born · age ~29
~21First venture · 2018
Self-reported$4.5B↗Series D valuation · 2026-01
Self-reported$250M↗Series D funding · 2026-01
Self-reportedEight figures↗Annual recurring revenue · 2024
Self-reported100+↗New enterprise customers · 2026-01

Key stages

What he was doing at each age

~21

Founded Lowkey as a student

~24

First startup exit

~26

Moved into enterprise AI for his second startup

~28

Second startup became a unicorn

Geographic moves

The city behind each change in direction

  1. United States
    1997 Born in the United States
  2. San Francisco
    ~21 Founded Lowkey as a student
  3. Lowkey
  4. Niantic
    ~24 First startup exit
  5. San Francisco
    ~26 Moved into enterprise AI for his second startup~27 Co-founder and CEO~28 Second startup became a unicorn~29 Co-founder and CEO

Selected work

An AI Agent platform for enterprises' chat, email, and voice customer-service workflows, focused on controllable workflows, evaluations, and continuous improvement

Lowkey

A recording and social-sharing product for gamers, acquired by Niantic in 2021

How the business works

Category
AI · Customer service Agent / enterprise software
Users and payers
Users are enterprise customer-service teams and end consumers; buyers are typically CX, operations, IT, and security departments
Pricing model
Charged by the volume of conversations handled by the Agent, with higher prices for complex tasks and voice; standard pricing is not public (Forbes report)
Main costs
Model and voice inference, enterprise integration, sales, and customer success, all of which grow with conversation volume
Validation level
Profitability unverifiedGross or net profit has not been disclosed; the business can retain profit only if the fee per conversation consistently exceeds inference and delivery costs
Moat
Workflow lock-in created by connecting ticketing systems, CRM, and internal APIs to handle refunds, cancellations, and account changes directly (analytical judgment)
Main risks
Competition from Salesforce, Zendesk, Intercom, and Sierra

Decagon's users include enterprise customer-service teams and end consumers, while buyers are typically CX, operations, IT, and security departments. Forbes reports that enterprises pay by the volume of conversations handled by the Agent, with higher prices for complex tasks and voice; the company has not published standard pricing. Forbes

Customers pay because the Agent can connect ticketing systems, CRM, and internal APIs to handle refunds, cancellations, and account changes directly, reducing manual processing costs. Model and voice inference, enterprise integration, sales, and customer success are the main costs, and they grow with conversation volume. The business can retain profit as long as the fee per conversation consistently exceeds inference and delivery costs; but Decagon has not disclosed gross or net profit, and its workflow lock-in still faces competition from Salesforce, Zendesk, Intercom, and Sierra.

Timeline

  1. Born in the United States

    Starting outUnited States

    Forbes reported in late 2025 that he was 28, confirming an approximate birth year of 1997; no reliable exact birth date was found.

    Source ↗
  2. Founded Lowkey while studying at Harvard

    About 21 years oldStarting outUnited States · Cambridge / San Francisco · Founder and CEO

    The product centered on recording and sharing gaming videos and building a player community—the first time he built a team and user network around a consumer product.

    Source ↗
  3. Lowkey acquired by Niantic

    About 24 years oldExitLowkey → Niantic · Founder of an acquired startup

    The acquisition price was not disclosed, so this records only the exit itself and does not conflate the company transaction with his personal proceeds.

    Source ↗
  4. Co-founded Decagon with Ashwin Sreenivas

    About 26 years oldTurning pointUnited States · San Francisco · Co-founder and CEO

    Zhang led as CEO and handled the business direction, while Sreenivas was the co-founder responsible for the early technical work; as of 2026, Decagon's website lists him as President, so he should no longer be described as the current CTO. Both entered the AI customer-service market with experience from their first startups and exits.

    Source ↗
  5. Came out of stealth and reached eight-figure ARR that year

    About 27 years oldMilestoneUnited States · San Francisco · Co-founder and CEO

    The company announced that it grew from zero to eight-figure ARR in 2024; this is a range, and this site does not turn it into an undisclosed exact figure.

    Source ↗
  6. Series C valuation reached $1.5 billion

    About 28 years oldMilestoneUnited States · San Francisco · Co-founder and CEO

    It took about a year to go from emerging from stealth in 2024 to becoming a unicorn, with customers including consumer internet companies and large enterprises.

    Source ↗
  7. Raised $250 million in Series D at a $4.5 billion valuation

    About 29 years oldMilestoneUnited States · San Francisco · Co-founder and CEO

    The official announcement said the valuation tripled in less than six months, and that the company added more than 100 global enterprise customers in the prior fiscal year.

    Source ↗

The First Startup Was Not Decorative Backstory

While studying at Harvard, Jesse Zhang founded Lowkey, a product for recording and socially sharing gaming content. In 2021, Lowkey was acquired by Niantic, the developer of Pokémon Go. The transaction price was not disclosed, but the exit itself gave him a complete experience of a company's product, team, and transaction cycle.

This experience makes Decagon different from the typical story of a first startup becoming a unicorn. Zhang and co-founder Ashwin Sreenivas had both founded and exited companies; for their second company, they chose an enterprise problem with more direct commercial value but heavier execution requirements.

Why Customer Service

Customer service has several properties that suit an Agent: high task volume, abundant text and voice data, outcomes measurable through resolution rate and satisfaction, and the ability for an Agent to execute refunds, cancellations, and account changes through APIs rather than merely answer questions.

But that also means mistakes directly damage customer relationships. Decagon's product focus is therefore not just fluent language; it lets enterprises define processes, connect systems, test performance, and keep improving.

Valuation Is Not Revenue

Decagon completed its Series C at a $1.5 billion valuation in June 2025, then raised at a $4.5 billion valuation in January 2026. The company officially confirmed only that it grew from zero to eight-figure ARR in 2024; Forbes reported a more precise annualized-revenue figure in late 2025, but that figure has no public financial statements behind it, so the list metric uses the official range.

These three numbers represent financing prices at two points in time and an operating run rate; they must not be combined into “the company makes $4.5 billion in revenue.” What remains to watch is whether revenue growth can outpace integration and service costs as enterprise deployments expand.

My read

Jesse Zhang is a serial founder who completed an exit with his first startup and carried his team and product experience into a larger market. It is easy to misread him as another person chasing the Agent trend, but Decagon chose an enterprise workflow with a clear budget, clear outcome metrics, and very high delivery difficulty.

The first lesson worth copying is to treat the first startup as a capability asset rather than an identity label: Lowkey's consumer product and community experience were not copied directly into customer service, but the experience of building a team, refining an experience, and completing a transaction remained. Second, Decagon does not only generate replies; it designs around workflows, permissions, tool calls, evaluations, and an improvement loop. Third, the two co-founders' commercial and technical responsibilities are distinct, but Sreenivas's current title is President, so the early CTO title should not be applied across the entire timeline.

The caution is that revenue from an enterprise customer-service Agent is not the same as high gross margin: deployment, integration, evaluation, and customer success can all be heavy. $4.5 billion is a financing valuation; the company only confirmed that it reached eight-figure ARR in 2024, and more precise figures subsequently reported by the media are not audited financial statements. Ordinary teams can learn to choose workflows with clear ROI, but should not ignore enterprise sales cycles and delivery costs.

Next founder