Vigil provides an AI defense platform that safeguards humans from unintended or malicious actions taken by AI agents across all digital surfaces, including finance, health, legal, and social domains. Its single Rust‑based engine operates uniformly on macOS, generating cryptographically sealed audit records and offering verifiable, revocable identities for each AI agent. By acting as an independent layer, Vigil prevents, revokes, and holds accountable AI‑driven actions that could otherwise cascade across services.
Funding
Funding not disclosed
Founders
Product
Problem
AI agents now act on behalf of users across finance, health, legal, and social domains without any independent monitoring, allowing unauthorized or harmful actions to cascade across services. Existing providers cannot audit themselves, creating a structural gap in accountability and protection.
Solution
Vigil provides an AI defense platform that intercepts every AI‑driven action across all digital surfaces. A single Rust engine runs identically on desktop, browser, and upcoming mobile/SDK deployments, inspecting requests in‑line and applying a deterministic policy engine. Each action is logged to a cryptographically sealed audit chain (VOAF) and the originating agent is assigned a verifiable, revocable identity. The platform can block risky actions, hold high‑impact operations for user approval, and instantly revoke tokens across all providers via a three‑layer emergency Kill Switch, delivering continuous protection and post‑event accountability.
Target Audience
Primary users are individuals and organizations that rely on AI agents for financial, health, legal, professional, and personal tasks, including enterprises seeking unified AI security across multiple providers.
Features
- TLS‑terminating proxy that captures all AI provider calls without requiring provider cooperation
- Four‑model detection ensemble (Isolation Forest, LSTM, Bayesian, Multi‑Window CUSUM) running locally with sub‑10 ms latency
- Cryptographically sealed audit records (VOAF) linked in a tamper‑evident chain for forensic and regulatory use
- Execution Gate that pre‑holds high‑risk actions for user approval and enables automated rollback where supported
- Emergency Kill Switch with instant local lockdown, OAuth/API key cascade, and network trust revocation
- Cross‑provider coverage and behavioral baseline building to detect anomalous actions across financial, health, legal, and other domains
- Single Rust codebase (11 crates) deployed across nine surfaces, ensuring consistent enforcement without forking