The RL ecosystem is maturing— verifiers are standardizing how we build and share environments. However, as it grows, we need observability tooling that actually understands RL primitives. Running RL experiments without visibility into rollout quality, reward distributions, or failure modes is a waste of time.

Monitor provides live tracking, per-example inspection, and programmatic access—see what's happening during runs and debug what went wrong afterward.

The future of RL observability could look like this:

You’re working alongside your model, spawning multiple versions of your environment by tweaking components at different points, much like using git worktrees for RL experiments.