Blockworks cuts crypto risk alerts from 10 minutes to 5 seconds with AI agents

In crypto, eight minutes might as well be eight hours. An exploit can drain a protocol, a bridge can pause, and a token can crater, all before a risk analyst finishes their first sip of coffee. Blockworks is betting that AI agents can close that gap almost entirely. The company launched “Agentic Detection” on September 17, a new feature inside its Monitoring product that uses proprietary AI agents to fire off risk and activity alerts in roughly five seconds. The previous workflow, which relied on manual analytical review, took eight to ten minutes. That’s a reduction of about 99%. What the system actually does The system pulls from X, Discord, Telegram, GitHub, governance forums, and more than 1,000 news sources. It covers over 1,000 assets organized into 55 subcategories, which gives it a fairly wide net for catching the kinds of events that make risk managers nervous: smart contract exploits, governance proposals, bridge pauses, and sudden shifts in project activity. When an AI agent flags something, the alert gets tagged as “Agent-detected” until a human analyst verifies it. That dual-layer approach is a deliberate design choice. Speed without accuracy is just noise, and institu...