Show HN: Running a vision model on every screenshot on-device

Category: other

Tags: ai-memory, privacy, screen-capture, gemma-4, rag, on-device, open-source

Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)

ScreenMind is a privacy-first, open-source AI memory tool that continuously captures and analyzes screenshots on-device using Gemma 4, enabling searchable, chat-accessible screen history with features like hybrid search, meeting transcription, and an agent platform. Its innovative combination of local multimodal AI, smart capture, and extensive integrations addresses privacy concerns raised by similar cloud-based solutions. The project is technically impressive with custom caching, GPU priority scheduling, and encryption, making it a compelling alternative for privacy-conscious users.

Target audience: backend devs, privacy-conscious users, power users, data engineers

Repository: https://github.com/ayushh0110/ScreenMind/blob/main/README.md · Python · MIT · 152 stars

View on Hacker News