A 95-year-old Harvard-educated Chinese American entrepreneur is leading a bold new venture aiming to redefine search for artificial intelligence — and has just secured $85 million in Series B funding, valuing the company at $700 million. The company, Exa, was founded in 2021 by Jeffrey Wang and his Harvard roommate, Will Bryk, who now serves as CEO. The recent funding round was led by Benchmark, with participation from Lightspeed, NVIDIA, and Y Combinator. Benchmark’s Peter Fenton will join Exa’s board of directors. Exa’s mission is clear: build a search engine not for humans, but for AI. As the company puts it, “Google search is for humans. We are for AI.” The idea emerged before the launch of ChatGPT in November 2022, when Wang and Bryk began developing a search infrastructure from scratch. Their early efforts included purchasing a GPU cluster to build a large-scale indexing system and experimenting with novel web search technologies designed to give AI agents the ability to navigate the web in ways humans cannot. The company launched its first product in November 2022, just two weeks before OpenAI unveiled ChatGPT. Soon after, Exa began receiving API requests from AI developers needing reliable, real-time web data. This marked a turning point: the team realized that AI, unlike humans, requires search at scale, speed, and precision — and that existing search engines were not built for this new reality. Exa’s search engine is purpose-built for AI agents. It features six core innovations. First, it prioritizes high-quality knowledge over SEO-optimized or ad-laden content, using a ranking algorithm designed to surface trustworthy, expert-level information. Because Exa does not accept ads, it avoids the incentives that degrade content quality in traditional search. Second, Exa delivers complete content for each result — not just a link or title — so AI models can process full context, such as entire articles, code snippets, or research papers, without needing to crawl pages manually. Third, speed is paramount. Human users tolerate delays, but AI agents make hundreds of search calls per task, and latency compounds quickly. Exa’s “Exa Fast” API achieves median latency under 450 milliseconds — significantly faster than wrapped Google search APIs, which typically exceed 700 milliseconds. In internal benchmarks, Exa achieved sub-50 millisecond latency from a U.S. West Coast data center. Fourth, Exa offers high computational power through its “Websets” product, designed for deep, asynchronous AI research. It enables AI to conduct comprehensive searches across vast datasets, such as global employee records, company profiles, or academic publications — making it the most exhaustive search tool available for AI applications. Fifth, the platform supports deep customization. Users can exclude thousands of domains, apply custom filters, and run tailored classifiers to align search results with specific AI workflows — a necessity for enterprise and specialized AI tools. Sixth, and perhaps most uniquely, Exa offers zero data retention. This means no user queries are stored on servers, in caches, or in logs — a critical feature for privacy-conscious enterprises. Unlike most search providers that route queries through Google’s infrastructure (and thus inherit Google’s data retention practices), Exa built its entire stack from the ground up. It crawls the web directly, trains its own models, and serves results from a custom vector database — ensuring that every query vanishes after processing. The company’s vision extends beyond current capabilities. Exa plans to expand its global index by five times and build a GPU cluster five times larger than its current setup. The goal: to organize nearly all publicly available information and create a foundational layer for future AI systems. Jeffrey Wang, a fluent Mandarin speaker, graduated from Harvard and previously worked at Plaid, where he built data and network infrastructure. His co-founder, Will Bryk, leads operations as CEO. The team includes several other prominent Chinese tech talents: Benjamin Chen (Harvard), Hubert Yuan (Tsinghua University’s prestigious姚班), Zixi An (Carnegie Mellon), Felicia M. Tang (UC Berkeley), and Benjamin Y. Chan (Cornell PhD). With backing from top-tier investors and a product uniquely tailored for the AI era, Exa is positioning itself not just as a competitor to Google, but as the essential infrastructure for the next generation of intelligent agents. Whether it can fulfill its ambition to surpass the world’s most dominant search engine remains to be seen — but the momentum is undeniable.