synstate labs

Human-AI Interaction Engineering

Synstate Labs is an independent research and engineering lab working on human-AI interaction. Our current focus is AI safety, AI dark patterns, and behavioral risks in long, personalized, and high-impact interactions.

Discover GuardContact
Abstract Synstate Labs field graphic
Current Work
Guard Synstate Guard evaluates AI-generated communication and actions before they reach the user. It detects AI dark patterns and other behavioral risks across the interaction and returns an explainable safety decision. We are currently developing the first version of Guard and preparing it for closed evaluation. Overview Live Demo Docs
Research
AI Dark Patterns ADP is a public registry of AI dark patterns and behavioral risks in AI systems. View Qualia Qualia is an on-device context model for AI agents. Its initial research phase explored how personal and work context can improve agent behavior. View