It's a modest launch, fresh ModernGuard MVP is out! --- Why: After working on Apple Siri and building internal guardrails, I realized that securing LLM applications — especially mission-critical ones — was a challenging task. Companies and developers face growing risks from prompt injections, jailbreaking attempts, and other adversarial attacks that could compromise AI integrity and business reputation (think fintechs, banks, legal, or healthcare companies). That's why I built ModernGuard, a fine-tuned modernBERT multilingual model that detects the latest prompt injection and jailbreaking techniques while delivering strong performance against benchmarks like PINT. But it goes beyond static defenses — it continuously learns from customer data and feedback, adapting to the context of each application, because context-aware security is essential for long-term protection. How: 🔹 Adaptive guardrails that learn from your feedback & data 🔹 A simple and fast API – with latency between 20 to 100ms 🔹 A console for monitoring logs, managing policies, and providing feedback (to update your model versions) My ask: I’d love your feedback. Check it out and try for free with 10M tokens: https://guardion.ai

ModernGuard LLM guardrail API

Fast and adaptive security control for AI agents

Fast and adaptive security control for AI agents