ML platform for experiment tracking, plus Weave for LLM/agent observability.
Category: Observability & Infra · Type: ML/LLM experiment tracking & observability · Website: https://wandb.ai
Overview
Weights & Biases is a leading platform for ML experiment tracking and model management; its Weave product extends this to LLM and agent applications with tracing, evaluation, and monitoring. It spans both training and agent-app lifecycles. W&B was acquired by CoreWeave in 2025 and operates as a CoreWeave company.
Key Features
- Experiment tracking and model registry
- Weave LLM/agent tracing and evals
- Dashboards and collaboration
- Strong fine-tuning/training heritage
Role in the Agentic AI Stack
Bridges model training and agent observability; overlaps with LangSmith and Langfuse on the LLM side.
Part of Agentic AI Landscape → Observability & Infra