Often the hardest part of programming, even when you are just programming along on your own computer, is understanding what your program is actually doing as it executes. Bugs that happen in production systems but cannot be readily reproduced in testing environments are the bane of software developers. 

What really makes bugs so difficult is that in order to track them down you need to know beforehand what kind of issue you are looking for so, you can log out and review the state causing the issue. But if you knew the issue beforehand you would likely already know how to solve the problem! 

Programmers are left putting logging everywhere or trying to infer state from other metrics they can observe (e.g. CPU load).

But what if instead of trying to figure out what to log beforehand we just captured all the states our software moved through all the time. We would be able to go back to any moment in time before a bug occurred and figure out what was going on. Irreproducible bugs and the guesswork that occurs when writing new code would go away — a capability that would be immensely valuable. 

This is precisely what Metawork is building. My only real question to them as an investor was, is this actually possible in a performant way? I have used early versions of the product and believe Metawork has proven that this functionality is possible and they will be the ones to deliver it. 

Today Metawork is launching the alpha of its always-on system that records every aspect of your code’s execution. We believe this has the potential to transform not just debugging production systems, but also how we learn and think about programming. As such I’m thrilled to share that they are also announcing $26M in funding, including a Series A led by Initialized with participation from CRV, and I will be joining their board.

It’s valuable to be able to stomp out hard to reproduce production bugs. What also seriously excites me is how transformative their approach may end up being to how we program and learn to program. The closest experience to Metawork’s level of insight is using a debugger, but those can be slow and tedious to use, and short of pair programming, almost impossible to use in collaboration with others. Metawork makes the same data easy to search and review and thus potentially very easy to share and collaborate on. I honestly can’t wait to see all the ways this will impact programmers’ everyday lives.

If anyone can do this, Metawork has just the team. Nick, David, Maggie, and David come most recently from Coinbase. Between them they have an immense amount of deep engineering experience paired with the vision and audacity to pull off such hard technology. 

Metawork is hiring. If hacking deep into VMs with the goal of changing how programmers think about and interact with their code sounds interesting to you, you’re just the kind of coder they’re looking for.