Scott Chacon helped build the platform that made Git ubiquitous. Now he's raising money to replace it.
GitButler, the Berlin startup Chacon founded about three years back, closed a $17 million Series A led by Andreessen Horowitz. Peter Levine, a general partner at the firm, is taking a board seat. Fly Ventures and A Capital, both early backers, came in for the round as well. The company announced the funding on April 8, 2026.
"We're not building some 'better git,'" Chacon wrote in a blog post accompanying the news. "We're building the infrastructure for how software gets built next."
It's a bold claim from someone who arguably did more than anyone—save perhaps Linus Torvalds himself—to cement Git's dominance. Chacon co-founded GitHub, authored the canonical text "Pro Git," and spent years evangelizing the tool his new company is now attempting to supersede. That he's back in version control, armed with venture capital and a thesis that Git is fundamentally outdated, says something about where he thinks the industry is headed.
When One Branch at a Time Isn't Enough
The pitch rests on a straightforward observation: Git was designed for an earlier era. It assumes developers work on a single feature, finish it, then move on. Branch, commit, merge. Repeat.
Modern engineering doesn't work that way. Teams toggle between multiple features simultaneously. Code review cycles stretch across days. And increasingly, AI agents spit out code changes faster than any human can shepherd them through the usual branch-and-merge ritual.
GitButler's answer is what it calls "virtual branches"—a system that lets developers apply multiple branches to a single working directory at once. Each gets its own staging area. Developers drag changes between them. The company frames this as "parallel work" rather than the endless context switching Git forces on you.
The client works with existing Git repositories, so teams don't have to abandon their history or retool their entire workflow overnight. It's built on Tauri with a Rust backend. There's also a CLI, somewhat cheekily named "but," that shares the same underlying engine. The project's GitHub repository has picked up approximately 20,300 stars, which—while not a perfect metric—suggests at least some appetite for what Chacon is selling.
Rebases That Don't Make You Cry
The technical details reveal where GitButler diverges most sharply from traditional Git workflows. Rebases, in GitButler's model, "always succeed." Conflicts get marked and deferred rather than blocking progress entirely. There's an undo timeline that logs every operation, which anyone who's accidentally blown away local work will appreciate.
Perhaps more significantly, the tool offers first-class support for stacked branches—a workflow increasingly popular at companies that ship code continuously. When you edit an earlier commit in a stack, GitButler automatically restacks everything above it. No manual rebasing, no cryptic error messages about detached HEAD states.
The company has also been integrating with AI coding tools—hooks for Claude and Cursor, an experimental "Agents" tab that ties coding sessions to specific branches, even a Model Context Protocol server that exposes GitButler actions to AI agents. These features were updated recently, though whether they'll prove essential or gimmicky remains an open question.
GitButler ships with integrations for GitHub and GitLab, allowing developers to open pull requests directly from the client. There's also a "Gerrit Mode" designed for commit-based code review workflows—the kind used by teams building Chrome or Android, where scale demands something more structured than GitHub's pull request model.
The CLI launched in technical preview in early February and arrived on Linux about six weeks later. It's available across macOS, Windows, and Linux on both x86_64 and ARM64 architectures.
The Team Behind the Bet
Chacon isn't alone in this. His co-founders include Kiril Videlov, who previously started Sturdy (a Y Combinator alum from the Winter 2021 batch) and Codeball, plus Anne Leuschner handling operations. It's a small team taking on an entrenched standard, which is either admirably ambitious or slightly insane depending on your tolerance for risk.
The a16z investment memo, co-authored by Levine and partner Matt Bornstein, leans heavily into the shift toward what Silicon Valley has started calling "agentic coding." Their argument: Git's single-branch index model collapses when AI agents start generating rapid-fire changes across multiple features simultaneously. That thesis maps neatly onto what GitButler has been building.
The product runs under a Fair Source license that converts to MIT after two years—a compromise between keeping some control and eventually going fully open. Pricing details haven't been made public. The downloads page offers installers but no visible payment tiers, at least not yet.
GitHub Flow, Reimagined
GitButler calls its approach "Butler Flow," essentially a rethinking of GitHub Flow for an age of parallel branches. The system enforces an explicit target branch and requires every diff to belong to a virtual branch. The idea is to surface merge conflicts earlier, enable continuous integration across team branches, and handle automatic cleanup once changes land upstream.
Whether this workflow catches on beyond the early adopters remains unclear. The funding announcement generated significant chatter on Hacker News—one thread drew 269 upvotes and well over 500 comments within days. That's a decent signal of interest, though interest and adoption are different things.
The competitive landscape is crowded. Established Git clients like GitKraken and Sublime Merge already serve developers frustrated with command-line Git. Newer experiments like Meta's Sapling and Google's Jujutsu are rethinking version control from different angles. And Cursor's acquisition of Graphite—a stacked pull request tool—last December suggests the broader industry is moving toward the incremental, stack-oriented workflows GitButler is betting on.
None of which guarantees GitButler will succeed. Version control tools are sticky. Developers learn Git once, grudgingly, and rarely want to learn something new. The fact that GitButler works with existing repositories helps, but asking engineers to fundamentally change how they think about branches is still a big ask.
The $17 Million Question
With the Series A closed, Chacon and his team now have runway to prove their thesis: that version control can evolve beyond the paradigm that powered his first startup's success. It's a curious position to be in—betting against the very tool you helped popularize.
Then again, perhaps that's exactly why he's the one to do it. Few people understand Git's strengths and limitations as intimately as someone who spent years helping millions of developers adopt it. If anyone has earned the right to argue Git's time is passing, it's probably the guy who wrote the book on it.
Whether developers agree is another matter entirely.
