Problem: You have a flamegraph. You don't know which function to optimize.
Solution: Drag any profile file here → get plain English explanation with specific fixes.
Supported formats
perf.data, pprof, JFR, .trace, speedscope.json, V8 (.cpuprofile/.heapsnapshot), HPROF, memray, Callgrind, collapsed stacks, systing, GHC, JSC heap snapshots.
What you get
- Top bottlenecks + line numbers
- Root cause (allocations, lock contention, cache misses, complexity)
- Concrete fixes: "replace
Vec::pushwithVec::with_capacity", "add index", "reuse buffer" - Estimated impact
Free
Bring your own OpenAI/Anthropic/Nvidia API key. No account, no tracking, I never see your key.
Try it with a profile from your last incident. Happy for feedback.



