Skip to content

Code to make any AI have unlimited context persistent memory. In the example, a software for any AI to read the Uniform Commercial Code of Michigan. A document of 220,000 tokens

Notifications You must be signed in to change notification settings

Daniel-codi/Concept_Curve_Embeddings_Indexation

Repository files navigation

Concept Curve Embeddings Indexation

A new algorithm that allow unlimited-context for any LLM, using Smart-Functions and Concept Curve theory to parse and reason over internal documents without RAG or embeddings retraining.


What is this?

This project demonstrates an alternative to traditional Retrieval-Augmented Generation (RAG), enabling AI to process internal documentation with:

  • Unlimited context memory
  • No fine-tuning or embeddings required
  • Concept Curve-based reasoning
  • Smart-Functions compatible with any LLM

Video Explanation

English Video showing the Demo working: https://youtu.be/8XhS3kaHKc8 Spanish Video showing the Demo working: https://youtu.be/0HRdeUPqqQ0

📂 Additional Resources

Find full documentation on Google Drive: https://tinyurl.com/CC-freeDocs

Full documentation & code for Spanish Users: https://tinyurl.com/CC-freeDocs contains the "Civil and Commercial Code of Argentina" in the spanish software

Find Articles as the work progresses: https://www.reddit.com/r/AIntelligence_new/

Find pre-print Paper: https://osf.io/preprints/osf/upm94_v1

English Paper: tinyurl.com/CCPaper-English Chinese Paper: tinyurl.com/CCPaper-Chinese

The preliminary version of the Concept Curve Paper has been released on May 11th, 2025, and is available in the Google Drive folder tinyurl.com/CC-freeDocs

Ask any questions you have to Agent-CC https://tinyurl.com/agent-cc


Features

  • Smart-Functions (AI-driven context tools)
  • Concept Curve Embedding Indexation
  • Context-aware prompt optimization
  • Lightweight document chunking system
  • Works with OpenAI and can be adapted for local LLMs

Getting Started

All the information is available on the /Documentation folder

Author

Daniel Bistman
Argentina 🇦🇷
Graduate in Business Administration, independent AI researcher. Inventor of Concept Curve paradigm for AI.

For more info, check the YouTube Channel - https://www.youtube.com/@Agente_Concept_Curve

Google Drive Resources - https://tinyurl.com/CC-freeDocs - (spanish) https://tinyurl.com/CC-docsGratuitos

(Paper Concept Curve Paradigm) https://drive.google.com/file/d/12ni96QekZB04SpGnNGXssHUF2DXDV-V8/view?usp=drive_link (Annex 3) https://drive.google.com/file/d/1lPtfeky60o0VgYURB4Q0nAwAZenyF4up/view?usp=drive_link ( CC论文_预发布版 ) https://drive.google.com/file/d/1pCQV3nUjdc41CtuMV9MYZykf1OVvC6ke/view?usp=drive_link

💬 Agent-CC - https://tinyurl.com/agent-cc

📧 agent.concept.curve@gmail.com - daniel.bistman@gmail.com


Donations (thanks in advance): 🅿️ paypal.me/DanielBistman

XRP Adress ---> rNFugeoj3ZN8Wv6xhuLegUBBPXKCyWLRkB

Memo required: 1868264810

XLM Adress ---> GAJ4BSGJE6UQHZAZ5U5IUOABPDCYPKPS3RFS2NVNGFGFXGVQDLBQJW2P

Memo required: 1868264810

📜 License

Free to use and distribute with attribution. If this project has helped your developer career, consider supporting the work.

Regards & blessings, Daniel Bistman

About

Code to make any AI have unlimited context persistent memory. In the example, a software for any AI to read the Uniform Commercial Code of Michigan. A document of 220,000 tokens

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published