- likes
- 123
- comments
- 6
Post
Anthropic just gave Claude Enterprise a control point to govern risky prompts before they ever reaches the model. Island now adds that governance. Inference Hooks route every governed prompt and tool result on claude.ai, Cowork, and Claude Code to an organization's own AI security server for an allow or deny verdict, before inference runs. That verdict now routes straight into Island's policy engine, the control plane for the agentic enterprise, the same layer already governing Claude traffic on the browser, network, APIs, endpoint, and everywhere work happens. No new browser, extension, or MDM profile required. One more surface accounted for. Same policy engine, same audit trail, same identity. New blog by Avishai Winiwarter https://lnkd.in/gc8f8ynv Roi Leibovich Tamar Cohen Alon B.