AI for Automation
Back to AI News
2026-03-26AI agentsstartup fundingOpenAIAI automationmulti-agent systems

He dropped out of Harvard at 23 — OpenAI just gave him $94M

Two 23-year-old founders just raised $94M at a $650M valuation to build software that coordinates thousands of AI agents. OpenAI quietly backed the deal.


Two 23-year-old AI researchers just convinced OpenAI to quietly invest in their startup — and it's already valued at $650 million. The company is called Isara, and its pitch is deceptively simple: build the software that lets thousands of AI agents work together on problems too big for any single bot.

Eddie Zhang, Isara's co-founder, dropped out of a computer science PhD at Harvard to build this. Before that, he spent time at OpenAI's safety team, where he studied how AI agents behave when you scale them up — how they coordinate, conflict, and occasionally break. Now he's turning that research into a product.

Isara AI agent coordination workflow diagram

What Isara actually builds

Right now, AI assistants like ChatGPT or Claude work solo. You ask one bot to do one thing. Isara wants to change that by building infrastructure where thousands of AI agents collaborate simultaneously — splitting up work, sharing data, and solving problems that would overwhelm any single model.

Think of it like this: instead of one intern doing everything, you get a trained team where each member handles a specific part of the job. One agent analyzes financial data. Another writes the report. A third double-checks the numbers. All coordinated automatically, without a human micromanaging each step.

The $94M question: Can you actually make thousands of AI agents work together without chaos? That's the hard part. Individual agents are powerful but unpredictable. Put a thousand of them in a room and you get collisions, conflicting goals, and wasted compute. Zhang's safety research at OpenAI was specifically about this problem — and now he's betting his company on the solution.

The team behind it

Zhang co-founded Isara with Henry Gasztowtt in San Francisco in June 2025. Both were 23 at the time. Since then, the tiny startup has recruited more than a dozen researchers from Google, Meta, and OpenAI itself — an unusual hiring spree for a company barely a year old.

Zhang's background is unusually deep for someone his age. He completed both a bachelor's and master's in computer science at UC Santa Barbara (with a 3.96 GPA), earned a Google Cloud Research Grant worth $87,200, and held a Harvard AI Safety Technical Fellowship before joining OpenAI's safety team. At OpenAI, he worked on systems that supervise complex model behaviors and early prototypes of multi-agent coordination.

Isara logo

Why OpenAI invested — and what it signals

According to the Wall Street Journal, OpenAI backed Isara at the $650 million valuation. This isn't just a check — it's a strategic signal. OpenAI builds individual AI models. But the future isn't one model doing everything. It's armies of specialized agents working in concert.

The use cases Isara is targeting include:

Financial analysis — Multiple agents simultaneously parsing earnings reports, market data, and news to generate real-time investment insights
Customer support automation — Agent teams that handle different types of queries, escalate edge cases, and learn from each other's interactions
Biotech research — Coordinated agents tackling different aspects of drug discovery or genomic analysis simultaneously

The bigger picture: from solo bots to AI workforces

Isara isn't alone in this space. Companies like Ruflo are building multi-agent systems for coding. ByteDance's Deer Flow just open-sourced a multi-agent research framework. But Isara is playing a different game — they're not building the agents themselves. They're building the operating system that makes agent armies possible.

If this works, the implications are significant. Today, companies might use one AI chatbot for support and another for coding. Tomorrow, they could deploy hundreds of specialized agents that coordinate automatically — handling entire business processes from end to end.

The $650 million bet is that two 23-year-olds who studied AI safety will be the ones to figure out how to make that safe and reliable. Whether they can deliver is the open question. But the fact that OpenAI — the company with arguably the most to gain from agent orchestration — chose to back them rather than build it in-house? That's the signal worth watching.

Related ContentGet Started with Easy Claude Code | Free Learning Guides | More AI News

Stay updated on AI news

Simple explanations of the latest AI developments