Astroware is an AI Alignment and Security research lab specializing in Constitutional AI. We build state-of-the-art guard models, safety classifiers, and adversarial evaluation frameworks that make AI systems safer by design. Our flagship products include Trishool, a decentralized agentic security runtime, and Petri, an adversarial red-teaming framework for LLM safety.
We also work directly with enterprises, AI labs, and governments as trusted advisors on AI safety strategy, model alignment, guardrail implementation, and regulatory readiness. Our team brings 20+ years of enterprise leadership from Oracle and the financial sector, combined with deep technical expertise in mechanistic interpretability, constitutional training, and decentralized AI infrastructure.

Astroware AI

Org chart

Teams

Offices