ly is a human-in-the-loop firewall designed to provide access control for AI agents, specifically managing and approving their network and file operations. It addresses the need for organizations and individuals to maintain oversight over what their AI agents can access, offering a system where requests for external connections or file access are intercepted and evaluated according to user-defined rules. If a request does not match existing rules, users are notified and can approve or deny access through channels such as Discord, Slack, Telegram, or email.
The platform supports both simple allow/deny lists and more advanced scripted rules using JavaScript, all of which are enforced locally on the user's infrastructure for low-latency evaluation. This local enforcement ensures that traffic never leaves the user's environment for rule evaluation, with the option to enable remote logging for audit trails when additional visibility is required. Users can grant time-limited permissions, create permanent rules, or deny requests with specific guidance that is relayed back to the agent. Audit logging is opt-in and can be configured for full visibility into agent requests and responses.
ly is available for Linux and Windows, with support for Mac planned in the future. The tool can be integrated into workflows via notifications on popular platforms like Discord, Slack, Telegram, and email, with SMS and app notifications coming soon. The agent runs alongside the AI system and does not require external calls for rule evaluation, ensuring privacy and control remain with the user. A cloud dashboard is available for certain plans, providing centralized management and alerts.
99 per month that includes cloud features and up to three protected machines, and an enterprise plan with custom pricing for unlimited agents, dedicated support, and advanced features such as SSO and SAML. All plans begin with a 7-day free trial that does not require a credit card. ly is positioned as an access control and observability solution for AI agent operations, providing granular and auditable oversight for both individual and organizational users.
In the LLM eval & observability space, watch.ly takes a focused approach. It focuses on securing and managing network and file access for AI agents with human-in-the-loop approvals. watch.ly is a B2B product aimed at AI developers and IT administrators. watch.ly follows a freemium model. watch.ly is available on the web, Windows, and Linux.
watch.ly first shipped in 2024. Among its 5 catalogued features are access control, approval workflows, and agent monitoring.
Latest indexed changes and source events
Other apps tracked under the same category.
