OpenAI backs Isara's $94M bet on AI agent swarms

What's the deal? OpenAI has backed Isara, a San Francisco AI startup developing software that can coordinate thousands of AI agents working in tandem on complex problems.

Founded in June 2025 by two 23-year-old researchers — Eddie ZhangDealroom has a profile for this one. Try Dealroom →, a former OpenAI safety researcher, and Henry GasztowttDealroom has a profile for this one. Try Dealroom →, then an OxfordDealroom has a profile for this one. Try Dealroom → undergraduate — the company raised $94 million at a $650 million valuation, with investors also including Michael OvitzDealroom has a profile for this one. Try Dealroom → and Stanley DruckenmillerDealroom has a profile for this one. Try Dealroom →.

Isara's core idea is to deploy large swarms of AI agents — potentially thousands — to tackle tasks too complex for a single model. It has already demonstrated the concept at an Allen & Co. conference in Arizona, coordinating roughly 2,000 agents to forecast the price of gold. Its first target market is financial services, selling predictive modelling software to investment firms.

Why now? Isara is part of a broader wave of so-called neolabs — research-focused AI startups founded by alumni of OpenAI, Google DeepMind, and other leading labs. The thesis: the incumbents have grown too large and too locked into existing approaches to explore new frontiers like large-scale agent coordination. Investors are rushing to back them before that window closes.

OpenAI's own involvement adds an interesting twist. The company is itself racing with AnthropicDealroom has a profile for this one. Try Dealroom → to sell agent-based tools to enterprise customers — yet it has chosen to back a startup directly competing in that space, suggesting it sees strategic value in staying close to the research.

What could go wrong? Coordinating thousands of AI agents in real time is an unsolved research problem. Isara has a team of roughly a dozen researchers and months of heavy compute spending ahead before it has a product. The graveyard of well-funded neolabs is already filling up: Adept, Inflection AI, and Character.AI all struggled to compete and ended up folding their talent into larger firms, with investors receiving modest returns.

At a $650 million valuation for a company less than a year old with no revenue, the risk-reward calculus is steep. Investor enthusiasm for founder pedigree may be outrunning commercial reality.

The signal: The AI agent race is attracting a new generation of insurgent labs betting that the next breakthrough will come not from scaling a single model but from orchestrating many. Isara, Humans&, and their peers are collectively drawing billions in investor interest — a sign that the frontier of AI competition is shifting from who has the biggest model to who can best coordinate armies of them.

Sources:
Tech Funding News
The Wall Street Journal
The Information
Republic World
Phemex

Image credit:
Isara

J.V.