obstruo.ai

About obstruo.ai

Obstruo is the sovereign AI control plane — a neutral governance layer that lets organizations govern what their AI can see, say, and do across every LLM, MCP server, and agent.It sits as a single drop-in control point in front of all AI traffic. Apps and agents connect using one Obstruo key and endpoint. Every request is automatically redacted, routed, enforced, and audited before it reaches any provider (OpenAI, Anthropic, Google, Mistral, Azure, AWS Bedrock, etc.).Key capabilitiesReal-time redaction & enforcement: PII, secrets, jailbreaks, and prompt injections are detected and blocked using regex, NER, and LLM-based methods tailored to GDPR, CCPA, and regional standards. Routing & kill switches: Policy-, cost-, or region-based routing with instant failover. Scoped kill switches for immediate hard kills, degradation, or quarantine. Unified observability & auditing: Complete, exportable audit trails that preserve redaction state and routing decisions for every call. Governance registries: Auto-discovery and risk classification of agents, MCP servers, and skills with per-tool permissions and framework readiness tracking (EU AI Act, GDPR, SOC 2, ISO 27001). MCP gateway: Govern tool access and permissions for Model Context Protocol servers. Obstruo is provider-neutral by design — we don’t build models, we govern them. It delivers EU data residency, ~20 ms gateway overhead, and in-path enforcement (not just monitoring), so compliance and engineering teams can deploy AI faster while maintaining full control and audit-ready evidence.Your AI. Your data. Your control. One neutral layer in front of every call.