AI Potluck
Product / UX / Telemetry & observability

MLflow

Databricks

Active 2026; Apache 2.0; observability/tracing for agents+LLMs+ML

Active 2026; Apache 2.0; observability/tracing for agents+LLMs+ML

Openness

5 high confidence
5.0
license
Apache-2.0
source
public
core-gated
ungated

license:Apache-2.0;source:public;core-fully-OSS(Databricks-Managed-MLflow is third-party hosting, not gated core)

Adoption

5 high confidence
5.0

Re-scored 4->5. 37.10M PyPI downloads in the last 30 days (verified on pypistats; the prior note already reported 25-30M+ but mislabeled the reach as 1M-10M). The map's prevailing usage_volume level-5 floor is >10M/mo (cf. pydantic-ai ~31M, langgraph ~58M at level 5). MLflow is the de facto open experiment-tracking / ML-lifecycle platform and clears the floor comfortably.

Capability

4 high confidence
4.0

Single open platform combining distributed tracing, prompt versioning, automated evaluation, and trace replay for agents/LLMs plus classic ML tracking. No standard benchmark for observability; rated on feature matrix.

Unchanged since 2026-07-30 (last edited, not re-checked)