ArthurAI

Arthur (incorporated as ArthurAI, Inc.) develops enterprise software for governing, monitoring and improving artificial intelligence systems. The platform originated as a machine learning observability product that automatically detects data drift — the divergence between the data a model was trained on and the data it processes in production — along with accuracy declines and fairness issues, surfacing them in dashboards with alerting so engineers can locate the scope and cause of a malfunction. It also supports comparison of competing model versions so teams can deploy the most reliable one. The platform is built as a collection of microservices designed for scalability and is documented as processing up to 1 million data operations per second.

The current product line is organized around what the company calls an AI control plane. On the governance and security side, Arthur offers agent discovery across OpenTelemetry, MCP, network and cloud APIs, risk scoring, ownership assignment and policy enforcement across large agent fleets, supported by SOC 2, RBAC, SSO, BAA and immutable audit logging. For engineering and product teams, an agent development toolkit provides prompt versioning with rollback, A/B experiments on production data, full-trace observability and live evaluations. Shared primitives include automated pass/fail AI performance evaluations, agentic observability and tracing across LLM calls, tool use and retrieval, and policies and guardrails that enforce the same evaluations synchronously before and after model calls. The company describes the platform as model-, framework-, cloud- and use-case-agnostic, covering traditional ML, generative AI and agentic systems. Flagship named offerings include Arthur Engine and Arthur Bench; both are published as open-source repositories on the company's GitHub organization, with Bench positioned as a tool for evaluating LLMs.

Founding story

ArthurAI, Inc. was established in 2018. Co-founders are identified as Adam Wenchel, David Nguyen, John Dickerson and Liz O'Sullivan, with Adam Wenchel serving as chief executive officer.

Business model

Enterprise software sold on a subscription basis, with tiered pricing plans matched to customers' AI deployment and management needs and transactions typically structured as annual contracts for use of the Arthur platform for model monitoring and evaluation. Professional services are offered to support enterprise-grade deployment.

Subscription revenue from annual platform contracts, with per-tier pricing; a free self-service entry point and open-source components are also offered.

Traction

The company reported growing annual recurring revenue by an average of 58% per quarter in the year preceding its September 2022 Series B and more than quadrupling revenue over that 12-month period, plus 235% recurring revenue growth in the first two quarters of 2022. Customer case studies on the company's site cite reduced model-monitoring maintenance workload and rapid time to first production model. Open-source traction includes 428 stars on the Bench repository and 88 on Arthur Engine.

Latest developments

Recent public activity centers on agent-era products: agent security and governance with automated agent discovery and policy enforcement, an agent development toolkit with versioned prompts and production A/B experiments, and content on operating production AI agents, FISMA-aligned governance for federal environments and EU AI Act compliance. Open-source repositories including arthur-engine, arthur-common, an Arthur Engine guardrails tool for agent frameworks, and custom-metrics documentation have seen updates into 2026.