He spent his Palantir years keeping California from burning. Now he is after a quieter fire: the paperwork that runs government.
Walk into a government procurement office and you will find some of the most consequential spending in the country held together with Microsoft Word, legacy software a decade or two out of date, and a handful of overworked civil servants who know where the bodies are buried. August Chen looked at that and saw the work of his life.
Chen is the co-founder and CEO of Hazel, an AI-native platform that helps government teams define what they need, write the solicitation, research the market, and evaluate the responses - all in one place. The promise is almost rude in its simplicity: a compliant first-draft solicitation built on the agency's own template in five minutes, not five months.
It sounds like a small thing. It is not. Roughly 30% of procurement roles in government sit unfilled as staff retire or quit, and the people left behind are responsible for buying everything from school buses to naval systems. Hazel is the extra pair of hands - the AI superpower, as Chen puts it - for teams short on staff and time but hungry to do the job better.
Since launching through Y Combinator's Winter 2024 batch, Hazel has gone on contract with agencies ranging from K-12 school districts to the U.S. Navy, and in July 2025 signed the City of Dallas as the first major Texas city to run procurement on AI. Not bad for a company whose founders chose, on purpose, the single hardest customer in technology.
The product itself is deliberately unglamorous in the best way. Civil servants can save legal-approved boilerplate, design content sections with Hazel's AI agents, and drag and drop them into the agency's own templates. They edit collaboratively in real time and export cleanly to Word when they are done - no copy-paste, no emailing documents back and forth, no cursory Google searches standing in for market research. Chen describes it as a consumer-grade tool that surfaces the critical information a human decision-maker actually needs, then gets out of the way.
Dallas is proving that good government and great technology go hand in hand.
Before procurement, Chen thought big. Really big. His Harvard thesis was on geoengineering - specifically, how you might safely deploy aerosols into the stratosphere to cool the planet. He graduated in 2020 with a degree in mechanical engineering and computer science, and along the way ran MakeHarvard, the university's student hardware hackathon.
That hackathon matters, because it is where he met Elton Lossner, the electrical-engineering classmate who would become his co-founder. The two went their separate ways after graduation - Chen to Palantir, Lossner to Boston Consulting Group - but kept the same compass heading: public service.
At Palantir, Chen built wildfire-prevention applications credited with a 99% reduction in acres impacted by wildfires in California, and his work helped utility PG&E save over $100 million in procurement costs. That last detail is the tell. Somewhere between the flames and the spreadsheets, he found the problem he could not stop thinking about.
Hazel did not start as Hazel. It started as Hazeltech - a marketplace helping small businesses find and win government RFPs. The logic was sound: bring more bidders to the table, governments get cheaper and better work.
Then Chen and Lossner did something most founders avoid. They admitted the model was indirect. If the mission was to help government, the fastest path was to sell to government - not to the businesses circling it. So they went direct B2G, knowing exactly what they were signing up for.
In their own words: long sales cycles, a litany of compliance standards, nobody returning your emails, and a whole new language of business. The kind of friction that sends most startups running toward easier money. Chen's read was the opposite - that the difficulty was the point.
Nothing worthwhile - or fulfilling - comes easy.
The pitch is not abstract. It is the difference between a quarter of a year and a coffee break. Drafting a single compliant solicitation, the old way versus the Hazel way:
Figures per Hazel's own product claims (YC launch). Bars are illustrative of the stated time gap, not to exact scale.
Hazel's small team is scattered across NYC, Boston, DC, and Salt Lake City, and reads like a who's-who of mission-driven institutions: alumni from Lockheed Martin, MITRE, SpaceX, Peraton, the U.S. Air Force, the State Department, and USAID.
Co-founder Elton Lossner, originally from Atlanta, spent about five years at BCG leading aerospace, defense, and public-sector cases before joining Chen full time. They have been building together since their hackathon days.
Hazel is a venture-backed Minority Business Enterprise (MBE), with early support from investors who previously backed Palantir and Anduril.
Chen frames procurement as a quiet lever on the country's biggest ambitions. Climate goals require rapidly buying clean-energy solutions. Great-power competition requires fielding new systems at scale. Aging procurement machinery, he argues, is the bottleneck on all of it.
When a maze of legal restrictions and understaffed offices makes it easier to re-up the incumbent than to recompete, the default vendor stops being a convenience and becomes an operational necessity. Hazel's bet is that better tools let government actually shop around again.
In July 2025, the City of Dallas became the first major Texas city to adopt AI for procurement, and Hazel was the partner. The arrangement is a clean test of Chen's whole thesis: take a large, fast-growing city straining under population growth, and hand its procurement teams a tool that compresses project timelines from months to days.
The platform automates RFP drafting and vendor identification across city departments. Just as important to Dallas, it widens the field - expanding opportunities for small and minority-owned businesses that have historically struggled to navigate the paperwork. For a venture-backed Minority Business Enterprise like Hazel, that alignment is not a marketing line; it is the founding premise carried forward from the company's marketplace days.
Chen does not talk about procurement as a back-office chore. He talks about it as the constraint sitting underneath the country's largest ambitions. Over $2.7 trillion flows through public procurement every year, and most of it moves through systems a decade or two behind their commercial equivalents, gated by incumbents who lean on political connections and litigation to stay on top.
That backlog touches everything from climate targets that depend on buying clean energy quickly, to national-security programs that depend on fielding new systems at scale. In October 2025, Chen carried the argument to the defense establishment directly, speaking at NDIA's event on modernizing the Department of Defense's requirements process - the same five-minutes-not-five-months pitch, now aimed at the Pentagon.
Chen's ambition is not a feature list. It is to change how governments buy and build at every level - faster, more transparent, more genuinely competitive - so that the institutions buying school buses and naval systems alike can actually keep up with what they are asked to do.
It is unglamorous work, and that is exactly the appeal. The hardest, least-celebrated corner of public service is precisely where he decided to plant the flag.