Arcade.dev
The secure action layer for production AI agents — authorized tool-calling at scale.
About
Arcade.dev, founded in February 2024 in San Francisco by Alex Salazar (CEO) and Sam Partee (CTO, ex-Redis), builds the authorization and governance layer that lets enterprise AI agents act in production. Its platform provides delegated per-user authorization at the moment of action, reliability engineering for tool calls, and a full audit trail of which agent did what on whose behalf — the gap that keeps most enterprise agent pilots from reaching production. The company authored the Model Context Protocol authorization specification adopted by Anthropic.
Arcade raised a $12 million seed in March 2025 (Laude Ventures) and a $60 million Series A in June 2026 led by SYN Ventures with strategic investment from Morgan Stanley and Wipro, totaling $72 million. It reports production use at a top US bank, Prosus, and LangChain, with tool-call volume up 25x in six months. Caveats: the 8,000+ tool count and 'only one in production at the world's largest banks' claims are vendor marketing, not independently audited.
What it can do
- Pre-built tools with delegated user authorization
- MCP-compatible tool runtime
- Full audit trail of every agent action
- Enterprise governance and compliance controls
- Self-host and air-gap options for regulated industries
Setup
Sign up at arcade.dev; install the Python SDK (arcade-mcp) and connect your tools.
Prerequisites
An Arcade account; OAuth credentials for the tools your agents will call.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 62; Trending score 78/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.