OCR
Optical character recognition; technology that converts images of text into machine-readable text, which performs poorly on complex historical layouts
As defined in Melissa Dell
Where it comes up · 7
- Max Simkoff - Redesigning the Home Buying Experience - [Founder’s Field Guide, EP. 47]2 claims
- Is Gemini File Search Actually a Game-Changer?2 claims
- Melissa Dell1 claim
- Feed Your OWN Documents to a Local Large Language Model!1 claim
- Eric Glyman - Reimagining Corporate Finance - [Invest Like the Best, EP. 275]1 claim
- Ai Sota #20261 claim
- This combo makes smart note systems magical.1 claim
Most important claims · 9 of 9Search all claims for “OCR” →
- Using multimodal LLMs (e.g., Claude Code with image input) to batch-OCR and summarize hundreds of PowerPoint slides exported as PNGs into a single markdown note delivers genuine 10x time savings — reducing roughly 4 hours of manual work to about 20 minutes of cross-checking.This combo makes smart note systems magical.
- Labs build OCR systems (DeepSeek OCR, OLMo OCR) because vast amounts of PDFs and digital documents on the web aren't easily encoded with text; extracting them yields what can be trillions of tokens of candidate data, which is described as synthetic data for pre-training and is a small percentage of the potential funnel that actually gets used.Ai Sota #2026
- Doma now uses computer vision to classify closing documents by layout and geographic placement of text, such as tables, shading and signature lines, to extract fees, reducing a task that takes a human about 30 minutes to under three seconds.Max Simkoff - Redesigning the Home Buying Experience - [Founder’s Field Guide, EP. 47]
- Card issuers never collected receipts, so employees forgot them and receipts arrived a month late, taking up to 24 hours to transfer from Amex to Concur. Ramp instead texts cardholders at the moment of purchase and uses OCR to match receipts, cutting average receipt turnaround to about 30 minutes.Eric Glyman - Reimagining Corporate Finance - [Invest Like the Best, EP. 275]
- Custom GPTs created through ChatGPT's 'Explore GPTs' interface can accept OCR'd PDF documents as part of their knowledge base, and these custom GPTs can then be shared via a link with access controls set to 'anyone with the link'.Feed Your OWN Documents to a Local Large Language Model!
- Standard tools like Amazon Textract and Google Cloud Vision produce garbage on historical data because they are geared toward single-column books and cannot handle complex multi-column newspaper layouts with headlines, images, advertisements, and captions; without manual digitization, researchers only have keyword counts and cannot reconstruct sentences, so they cannot extract sentiment or how communities discussed events—a gap Dell is closing with computer vision and NLP methods to unlock massive historical datasets.Melissa Dell
- Gemini File Search does not preserve document structure during OCR and text extraction—markdown headings are converted to plain text separated by newlines, losing the document hierarchy that is important for RAG retrieval quality.Is Gemini File Search Actually a Game-Changer?
- Doma's escrow inputs are often flat image PDFs, so it applies OCR to extract text, then natural language models trained to understand the syntax title professionals use, such as how vesting language implies tasks.Max Simkoff - Redesigning the Home Buying Experience - [Founder’s Field Guide, EP. 47]
- Gemini File Search can ingest many different file formats and can perform OCR on scanned PDFs and documents, extracting text from non-machine-readable images.Is Gemini File Search Actually a Game-Changer?