Reproducing rare bugs found during fuzz testing is hard because software can run slightly differently every time. Bedrock is a hypervisor that controls that randomness, so given the same inputs a program runs identically on every execution.
Niklas outlines an approach and a solution, bedrock, for this problem of determinism. He focuses on the challenge of emulating time and pausing a VM at an exact instruction count, something CPUs aren't designed to do.
https://brink.dev/blog/2026/06/25/bedrock-deterministic-hypervisor/
The result is an open source tool, bedrock: an experimental hypervisor purpose-built for deterministic software testing.
https://github.com/oss-garage/bedrock
There's a Jane Street podcast with the Antithesis founder that has a nice introduction to/motivation for deterministic hypervisors (it led me to the blog post Niklas links to in their post): https://www.youtube.com/watch?v=F_LvzcdNH3Q.
Yas queen. Also a reminder, to me, of how fun/powerful systems level programming is.