Arthur AI

Arthur builds a platform for evaluating, monitoring and governing AI systems, covering traditional machine learning models, generative AI and AI agents. Its offering includes pre-production and runtime evals, guardrails, step-level agent tracing, token and cost tracking, and agent discovery with runtime security. The Arthur Evals Engine is available as open source, and the company has released Deckard, an open-source scanner for rogue AI agents. The platform runs an evals engine in the customer's environment, so inference data stays local and only anonymised metrics reach the control plane.

Customers are enterprise AI teams, security, audit and governance teams, and product and engineering leaders, including in regulated industries such as banking, healthcare and insurance. Revenue comes from tiered subscriptions: a free plan, a Premium plan at $60 per month, and an Enterprise plan with custom pricing that adds dedicated or self-managed VPC deployment, SSO and SLAs.

The company's CEO is Adam Wenchel. Its open roles are listed as based in New York City and Washington, D.C., and the careers page describes hybrid work. Current openings are an AI Security Product Manager and a Forward Deployed Engineer.

Product: AI evals, monitoring, guardrails and agent discovery/runtime security platform, plus the open-source Arthur Evals Engine and Deckard scanner.