Tech Share: Genius AI Brain  <!-- .element: width="80%" --> --- ### Introduction - What is Brain Service? - What are Large Language Model (LLMs) and how to use them? - Development Plan for the Brain Service --- #### What Brain Service? - Focus is on domain experts. - Enhance interaction with AI or LLMs to make improvement.  <!-- .element: width="40%" -->  <!-- .element: width="40%" --> --- #### Brain Service requirements - Requires easy maintenance. - Strong support from the AI community. => Python and [Langchain](https://www.langchain.com/)  <!-- .element: width="60%" --> --- #### What Can We Expect from the Brain Service? - Stateless Design (mostly) - Experts easily changing the prompt and workflow - AI response records and statistics  <!-- .element: width="60%" --> --- #### The Brain Service - Dedicated platform for AI enhancement - Streamline development - Improves the accuracy and quality of AI output. - Focus on collaboration with domain experts  <!-- .element: width="40%" --> --- ### What are Large Language Models (LLMs)? - Introduction to LLMs - Prompting Strategy - Database Integration --- #### Introduction to LLMs - Core function: understand and generate human-like text or code output - Simple principle: text in, text out - Multimodal capabilities: text, images, audio, video  <!-- .element: width="80%" --> --- #### Prompting Strategy - Agents: autonomous entities performing tasks based on prompts  <!-- .element: width="80%" --> --- #### Prompting Strategy - Chain of Thought: prompting model to explain reasoning step-by-step - Tree of Thought: exploring multiple reasoning paths for solutions  <!-- .element: width="80%" --> --- #### Database Integration - Enhancing LLM functionality with database integration - Vector Databases: high-dimensional data handling for similarity search and recommendations - Graph Databases: managing complex relationships for improved contextual understanding  <!-- .element: width="40%" -->  <!-- .element: width="40%" --> --- #### Usages for the Project - Prompting - Code Generation - Workflow Automation - Error Checking and Exploration --- #### Usages for the Project - Knowledge Base - Manage data and Retrieval Information - Question Answering - Self Auto Improve --- ### Development Plan Overview - Outline the stages for developing the Brain Service - Focus on how to improve AI and workflow --- #### Stage One: Initial Workflow - Domain Experts: QA team - Collaboration for insights and feedback - Integration with [Landpress](https://landpress-content-v2.linecorp.com/projects/mmn2ni86hi6muysnargzn18t/content/collections/test_case_prompts/items) Content Service - Easy for changing the application settings and prompt improvement - Observation Tool: [Langsmith](https://www.langchain.com/langsmith) - Monitoring and analyzing AI interactions --- #### Stage Two: Quality Improvement - Development of a framework for optimizing prompting strategies - Custom workflow for domain experts to manage the pipe line - Creation of agent tools to automate tasks and improve AI's capabilities --- #### Stage Three: RAG (Retrieval-Augmented Generation) System - Implementation of a knowledge base using vector or graph databases - Enhancement of data storage, retrieval, and contextual understanding - Empowerment of AI to deliver accurate and insightful user responses  <!-- .element: width="30%" --> --- Thank you!