Sahil Lavingia didn't just walk away; he got the boot. Honestly, if you were following the headlines in early 2025, the story of the Gumroad founder joining Elon Musk’s Department of Government Efficiency (DOGE) felt like a Silicon Valley fever dream. A high-profile "minimalist entrepreneur" moving to D.C. to fix the Department of Veterans Affairs with a MacBook and some Python scripts.
Then, 55 days later, it was over. No big ceremony. No transition plan. Just a revoked GitHub access and a "ghosting" that would make a bad Hinge date look professional.
The Sahil Lavingia doge exit is a weirdly perfect window into why "disrupting" the government is so much harder than building a SaaS app. Sahil went in expecting to find a mountain of waste and easy wins. He left admitting the government was actually... kinda efficient? That confession is likely what cost him the gig.
The Interview That Triggered the Sahil Lavingia Doge Exit
You don't usually get fired from a volunteer role, but Sahil managed it. The catalyst was an interview with Fast Company where he broke the first rule of the Musk era: don't humanize the "bureaucracy."
In that interview, Lavingia was surprisingly candid. He told the reporter that the culture shock wasn't the laziness he’d been told to expect, but rather a world of endless meetings where few decisions were actually made. "It's not as inefficient as I was expecting, to be honest," he said. For a group like DOGE, which was predicated on the idea that the federal government is a burning pile of waste, that kind of nuance was heresy.
The timeline of the Sahil Lavingia doge exit moved fast:
- Early May 2025: The Fast Company piece goes live.
- May 9, 2025: Sahil realizes his logins are dead.
- The Aftermath: He publishes a "diary" on his personal blog, DOGE Days, laying out the reality of his 50-odd days in the trenches.
Munchables and Hallucinating AI
While at the VA, Sahil wasn't just sitting in meetings. He was writing code. Specifically, he built a tool designed to "munch" through thousands of federal contracts to find savings. He used Large Language Models (LLMs) to scan about 76,000 contracts, tagging them as "munchable" if they looked like waste.
It sounds efficient. It was also, according to some VA staffers, a bit of a disaster.
Congressional testimony later revealed that the "munchable" script had a habit of hallucinating. It misread contract values—seeing a $35,000 contract and thinking it was worth $34 million. It flagged essential services, like cancer research and nursing care, for termination. Sahil later defended the work, saying he never intended for the code to run autonomously. He compared it to an episode of The Office where Michael Scott drives into a lake because the GPS said so. "Do not drive into the lake," he warned.
But in the high-stakes environment of the VA, where those contracts provide literal lifelines for veterans, "vibe-coding" through the budget didn't win him many friends among the career civil servants.
✨ Don't miss: Verizon Phone Claim Insurance: What Most People Get Wrong About Repairs and Replacements
Why the "Silicon Valley" Approach Failed
The Sahil Lavingia doge exit highlights a massive friction point. In tech, you "move fast and break things." In government, if you break the "things," people don't get their disability checks or their chemo.
Sahil found himself hamstrung by a government laptop that couldn't run basic tools like Git or Python without a mountain of approvals. He wanted to open-source the VA’s code, a move he pushed for during an all-hands meeting with Elon Musk. While Musk liked the transparency angle, the actual implementation was a bureaucratic nightmare.
The Power Vacuum
One of the most insightful things Sahil noted after his exit was that DOGE didn't actually have any power. They were essentially "McKinsey volunteers." The real decisions were being made by Trump-appointed agency heads. DOGE was the "fall guy"—the group that got blamed for the unpopular layoffs and cuts while the political appointees held the actual reins.
Key Lessons from the Sahil Lavingia Doge Exit
If you're looking for the "so what" in all of this, it's about the limits of external disruption.
- Transparency is a double-edged sword. Sahil took the "radical transparency" of his company, Gumroad, and tried to apply it to a government task force. It got him fired in under two months.
- The "Easy Wins" don't exist. Most of the "waste" in government is locked behind laws, veteran preferences, and seniority rules (RIF rules) that a script can't just bypass.
- Code isn't a silver bullet. Building a "VAGPT" chatbot (which Sahil did) is great for UX, but it doesn't solve the underlying data silos that make claims processing take 130+ days.
The Sahil Lavingia doge exit wasn't a failure of talent, but a collision of worlds. He went in as a volunteer, hoping to use his skills for the public good, and left as a cautionary tale for any tech founder thinking they can "fix" D.C. in a weekend hackathon.
Actionable Insights for Tech Pros:
If you're looking to transition into "Public Interest Tech," don't go through a political task force. Look at established groups like the U.S. Digital Service (USDS) or 18F. They understand the "administrative state" well enough to actually ship code to production, rather than just getting ghosted on GitHub after two months of work.