He built an open-source tool AI coding assistants couldn’t ignore

Safi Shamsi built Graphify alone over a weekend, gave it away for free, and somehow ended up with AI coding tools leaning on it. The software now helps power Cursor, Claude Code, Codex, Gemini CLI and more than fifteen other assistants by doing one job those systems all need: it reads a codebase, maps how the pieces connect, and hands that map back to the tool asking the question.

The usage numbers are hard to ignore. Developers pull it down about 65,000 times a day, it hit 79,000 on its busiest day, and the running total has moved past 4 million since April. Graphify has also racked up more than 100,000 GitHub stars, though Shamsi seems much more interested in the harder-to-fake signs: more than 10,000 forks, a number-three global trending spot on GitHub in July, over a million mentions of the keyword on X in a month, and a stream of downloads that keeps climbing.

That traction got him into Y Combinator’s Summer 2026 batch without a cofounder, which is exactly the sort of setup the accelerator usually tells people not to bring. Jared Friedman, a YC group partner, said he was struck by what Shamsi had done as a one-person team and framed him as proof of the “100x or 1000x engineers” idea. The harder question now is different: Graphify Labs has staff, and the job has shifted from shipping the thing to running a company around it.

The defense against the obvious clone risk is that Graphify sits in a neutral layer. Any AI assistant needs a map of the code, and none of them wants to depend on a rival for that map. Shamsi says a bigger company can copy the product, but not the research he has already done. A former solo founder and investor, Gabriel Jarrosson, says the distribution is already the real story, and a Rootly plugin that uses Graphify on incident history is one sign other teams are treating it like infrastructure.

Shamsi is now pushing a second product, a Rust engine called Penpax for enterprises, with more than 4,500 names on the waitlist and, he says, Fortune 500 teams among them. The business is still ahead of him, but the pattern is familiar: open-source starts as a free utility, then becomes the layer everyone else quietly builds on. That’s the dangerous part for incumbents, and the interesting part for everyone else.