Many teams need to extract text from PDFs and scans but can't send sensitive documents to cloud services. Existing local tools produce messy output.
Package state-of-the-art OCR and layout analysis into a simple Docker container with a REST API. Focus on quality for technical documentation and legal texts.
Initial version could use existing open-source libraries wrapped in a clean interface. Later add fine-tuned models for specific verticals.
Primary challenge is competing with free cloud tools from big tech companies.