GlassLM

We built GlassLM after realizing how often we paste sensitive or confidential text into AI tools without knowing what actually gets sent or logged. Instead of asking users to “trust us,” GlassLM shows exactly what the AI will receive before anything is sent. Sensitive data is masked locally, and only placeholders reach the model. This launch focuses on transparency first, no accounts, no stored prompts, and full user control. We’d love feedback, edge cases, or thoughts on where privacy-first AI should go next.