Welcome - Doc-Vision.com - Blog
Welcome to our blog where we share updates, insights, and stories about document intelligence
DocVision: Your Financial Documents, Finally Under Control
A practical walkthrough of DocVision - the Financial Doc Agent. Upload invoices and bank statements, reconcile them, and build a custom expense report with Claude Code.
Run GLM-OCR Locally with Ollama - Free, Local, Ready in Minutes
GLM-OCR is a 0.9B vision-language model ranked
Why your agent reads PDFs wrong - and what to do instead
Why raw PDFs break LLM workflows, why PyPDF and screenshot-every-page fall short, and how DocVision layers structured extraction plus deterministic apps on top.
The Ultimate Guide to Automated Invoice Capture
Learn how AI-powered invoice capture eliminates manual data entry, reduces errors, and speeds up your AP workflow with Vision OCR+ and machine learning.
N8N, Google Drive, and DocVision Workflow
Build a complete automation that pulls invoices from Google Drive, extracts data using AI with DocVision, and automatically updates your Google Sheets - all without lifting a finger.
Engram - DeepSeek's New Axis of Sparsity Separating Memory from Reasoning
DeepSeek's Engram introduces conditional memory as a new axis of sparsity for LLMs. A modernized N-gram lookup table runs on the CPU, relieves early transformer layers from static recall, and unlocks big gains in reasoning, math, and long-context tasks.
Invoice Scanning - Go Paperless Painlessly
Learn how DocVision's OCR+ technology automates invoice scanning and eliminates manual data entry
Bayesian - AI That Evolves In Real Time By Google
Discover how Google researchers solved one of AI's biggest weaknesses - teaching models to learn and adapt in real-time using Bayesian reasoning, achieving 80% alignment with optimal strategies.
The Monty Hall Problem - When Intuition Fails
Should you switch or stay? Why almost everyone gets it wrong on the first try.
Is AI Smarter Than a Mosquito?
LLMs can seem remarkably capable in some settings and surprisingly limited in others. Explore the paradox of knowledge vs. generalization.
Run Gemma 4 26B MOE Locally on a Mac with Only ~6GB RAM
A step-by-step guide to running Google's Gemma 4 26B Mixture-of-Experts model locally on Apple Silicon using llama.cpp, mmap, and Metal - achieving 49 tok/s with under 6GB of RAM.
Building an AI agent that calls APIs? Read this first
The 3 hardest technical problems I hit building an AI agent that calls real APIs — and the fixes that actually work.