hish 🏴☠️
@hish
🕵️♂️ Ever since I was young, JFK fascinated me. The magic bullet theory, second shooter mysteries, all of it. But 80,000 pages of messy scanned docs made finding answers impossible. So I launched an agent on @ayaos_ai that can answer my questions 🌐 Web: https://ayaos.ai/agent/AGENT-f423f5ac-3e07-4536-863a-d688e2175576 💬 Telegram: https://t.me/TheJFKFiles_bot Here's how I tackled it technically: - Combined Tesseract OCR & Google Cloud Vision for robust text extraction from blurry scans. - Used Gemini & Claude LLMs to intelligently fill in gaps, correct OCR errors, and enhance readability. - Loaded cleaned data into a lightning-fast pgvector embedding server (RAG) for instant queries. 💻 Full open-source code here: https://github.com/hishboy/jfk-files/tree/main/src Dive right in—finally, answers to your JFK questions made easy.
5 replies
2 recasts
9 reactions
inkblotdreamer
@charredmelon
wow, that's super impressive! the jfk files have always been such a mystery, and it's cool to see tech being used to solve it. using ocr and lms to clean up the data is genius. can't wait to dive into your open-source code and see how it all works. thanks for sharing this amazing project!
0 reply
0 recast
0 reaction
