Runsight is a YAML-first workflow engine for AI agents that lets teams design agent workflows in YAML, commit changes to Git, and track the cost per run. It provides governance with built-in assertions to validate behavior, while supporting open-source, self-hosted deployments for privacy and control and easy integration into existing CI pipelines. The YAML-first approach keeps workflows readable, auditable, and portable across environments, so data scientists, ML engineers, and product teams can collaborate more effectively, reproduce experiments, and monitor spending as workloads scale.

Key features include:

• YAML-first workflow design: Create, version, and orchestrate AI agent workflows in clean YAML syntax, enabling fast iteration, reuse, and cross-project consistency.
• Git-backed collaboration: All changes are committed and traceable via Git, supporting reviews, branching, and reliable rollback in team environments.
• Cost tracking per run: Monitor per-run expenses with clear metrics, dashboards, and alerts to help teams budget, optimize runtimes, and justify resource use.
• Built-in assertions for evaluation: Validate outcomes with configurable checks, catch regressions early, and gate deployments with confidence.
• Self-hosted deployment: Deploy on your own infrastructure for privacy, control, and scalable on-prem usage, with flexible hosting options.

Runsight provides a robust, auditable platform for orchestrating AI agents at scale, balancing innovation with governance and cost discipline.