Screenshot Séance: I Built a Local AI to Find the Screenshots That Still Need Me

Description and use

Screenshot Séance is a local tool that uses Gemma 4 via Ollama to automatically classify screenshots into five buckets, streamlining task management and removing clutter.

Problem

Users accumulate screenshots that contain actionable items, memes, or sensitive data, but lack a way to triage them efficiently.

Outcome

Screenshot Séance automatically triages screenshots, reducing manual effort and ensuring sensitive items are flagged for review. The local Gemma 4 model runs without external API costs, and the application supports resumable processing and reversible file moves.

Workflow

• Scan folder and record metadata
• Detect near‑duplicates with perceptual hashes
• Send each screenshot to Gemma via Ollama /api/chat
• Validate JSON output and retry once if invalid
• Apply local rules for sensitivity, deadlines, and confidence
• Persist progress in SQLite for resumability
• Generate actionable list and calendar export
• Move eligible files to _released_by_seance folder
• Provide undo functionality

Evidence

Séance uses Gemma 4 running locally through Ollama to read a screenshot folder and sort it into five buckets

Source

The application code is MIT licensed. The repository includes the backend, frontend, synthetic screenshots, benchmark script, tests, and browser demo implementation.

Source

In an initial benchmark of ten real screenshots on a laptop with an RTX 2060, 6 GB VRAM, and 32 GB RAM, I compared two models with thinking disabled and images capped at a 1024-pixel longest side

Source