TokenTrim: screenshot and PDF compressor for LLMs

11/10/2026 23:00AIDevToolsProductivityRAG

Turns screenshots and PDFs into compact Markdown so RAG and agent users save tokens

Problem: Developers dump screenshots and PDFs straight into AI, wasting tokens and hurting RAG results, often because of file handling rather than the framework. Most OCR tools only recognize text and ignore whether output is compact or LLM-friendly.
Solution: Convert screenshots and PDFs into structured, de-duplicated Markdown: keep heading hierarchy and tables, merge fragmented paragraphs, trim to a token budget. Offer a local batch CLI and simple web UI with before/after token comparison.
Users: Indie developers building RAG/agents, small AI teams, and heavy users who ask AI questions with screenshots
MVP: Screenshot/PDF to Markdown; token-budget trimming and chunking; before/after token comparison; batch processing and export; local mode with no cloud upload
Revenue: Free tier plus $9-29/mo subscription; metered CLI batch processing; team license for private deployment
Why now: Multimodal input is exploding and token cost is a daily pain, while generic OCR ignores the 'feed the LLM' format problem; a solo developer can ship and validate quickly