diff --git a/AI_learn/LLM_Models_Locally_lmstudio.md b/AI_learn/LLM_Models_Locally_lmstudio.md
new file mode 100644
index 0000000..1142733
--- /dev/null
+++ b/AI_learn/LLM_Models_Locally_lmstudio.md
@@ -0,0 +1,274 @@
+## 1. Introduction to Running AI Models Locally
+
+* The video explains how to run **AI models (LLMs)** on a personal computer privately. 
+* **LLM (Large Language Model)** means AI models like ChatGPT-style models.
+* Two popular tools mentioned:
+
+  * Ollama
+  * LM Studio
+
+### Important Idea
+
+* Running models locally gives:
+
+  * **Privacy**
+  * **Offline usage**
+  * More control over AI models. 
+
+---
+
+## 2. AI Application + LLM Concept
+
+* Any **AI application** or **AI agent** needs two things:
+
+  1. The AI application itself
+  2. An LLM connected to it. 
+
+### Examples of AI Applications
+
+* GitHub Copilot
+* Claude
+* OpenCode/OpenClaw-like AI coding tools. 
+
+### Key Idea
+
+* First install the **AI tool/agent**
+* Then connect it to an **LLM**. 
+
+---
+
+## 3. Two Ways to Use LLMs
+
+### A. Cloud Hosted LLM
+
+* AI models hosted on cloud servers.
+* Accessed using:
+
+  * Website login
+  * API key. 
+
+### Examples
+
+* OpenAI
+* Anthropic
+* OpenRouter. 
+
+### B. Local LLM
+
+* Models run directly on your PC.
+* Tools used:
+
+  * LM Studio
+  * Ollama
+
+### Important Idea
+
+* Local LLM = More privacy and local control. 
+
+---
+
+## 4. Installing LM Studio
+
+### Steps
+
+1. Visit the LM Studio website.
+2. Download the installer for your OS.
+3. Run the installer.
+4. Accept agreement and install. 
+
+### Important Idea
+
+* LM Studio allows running many open AI models locally. 
+
+---
+
+## 5. Hardware Requirements
+
+LM Studio shows:
+
+* CPU information
+* RAM size
+* GPU details. 
+
+### Example from Video
+
+* RAM: 16 GB
+* GPU VRAM: 4 GB NVIDIA GPU. 
+
+### Important Terms
+
+* **GPU** → Graphics processor used for AI acceleration.
+* **VRAM** → GPU memory used for AI models.
+
+### Key Idea
+
+* Bigger models need stronger hardware.
+
+---
+
+## 6. Developer Mode and llama.cpp
+
+* Enable:
+
+  * GPU offloading
+  * Developer mode. 
+
+### Important Term
+
+* **llama.cpp**
+
+  * Runtime engine used internally by LM Studio. 
+
+### Key Idea
+
+* LM Studio uses llama.cpp behind the scenes to run models efficiently.
+
+---
+
+## 7. Downloading AI Models
+
+### Example Models Mentioned
+
+* Gemma4
+* Nemotron
+* Qwen. 
+
+### Steps
+
+1. Open model search.
+2. Select model.
+3. Click download.
+4. Load the model into chat. 
+
+### Important Term
+
+* **GGUF**
+
+  * A model file format optimized for local inference. 
+
+---
+
+## 8. Running the Model
+
+* After loading the model, user can chat with it directly. 
+
+### Features Mentioned
+
+* Thinking capability
+* Vision capability. 
+
+### Example
+
+* Gemma 4 responds to questions locally on the PC.
+
+---
+
+## 9. Tokens and Context Window
+
+### Important Terms
+
+#### Token
+
+* Small chunks of words processed by AI.
+
+#### Context Window
+
+* Maximum amount of text the model remembers in one conversation. 
+
+### Example
+
+* Context window shown:
+
+  * 4096 tokens. 
+
+### Key Idea
+
+* More conversation = more token usage.
+* When context limit fills up, older information may be forgotten.
+
+---
+
+## 10. KV Cache Explanation
+
+### Important Term
+
+* **KV Cache**
+
+  * Memory optimization technique used by LLMs. 
+
+### Key Idea
+
+* KV cache improves speed by remembering previous calculations.
+
+---
+
+## 11. Stopping Model Generation
+
+* User can stop the model while it is generating output. 
+
+### Key Idea
+
+* Useful when:
+
+  * Response is too long
+  * Output is unnecessary
+  * Model becomes slow
+
+---
+
+## 12. Choosing Models Based on Hardware
+
+### Important Idea
+
+* Large models require:
+
+  * More RAM
+  * More VRAM. 
+
+### Example
+
+* Gemma 4 27B model requires around 18 GB VRAM.
+* Smaller model (E4B) is suitable for weaker PCs. 
+
+### Key Lesson
+
+* Choose models according to your PC specifications.
+
+---
+
+## 13. Connecting Local LLM to AI Agents
+
+### Steps
+
+1. Enable developer mode.
+2. Enable local server/integration.
+3. Use local URL:
+
+   * `127.0.0.1:1234` 
+
+### Key Idea
+
+* AI coding tools can connect to local LLMs through this API endpoint.
+
+### Example Use Case
+
+* Connect local model to:
+
+  * Coding agents
+  * AI assistants
+  * Automation tools
+
+---
+
+## 14. Final Core Takeaways
+
+### Most Important Ideas from the Transcript
+
+* AI applications need an **LLM backend**.
+* LLMs can be:
+
+  * **Cloud hosted**
+  * **Locally hosted**.
+* LM Studio helps run AI models privately on a PC.
+* Hardware decides which model you can run.
+* Local models can integrate with AI coding agents using localhost APIs. 
diff --git a/AI_learn/opencode_AI_agent.md b/AI_learn/opencode_AI_agent.md
new file mode 100644
index 0000000..0f6df30
--- /dev/null
+++ b/AI_learn/opencode_AI_agent.md
@@ -0,0 +1,383 @@
+# OpenCode AI Agent Tutorial – Beginner-Friendly Summary
+
+
+---
+
+# 1. What is OpenCode?
+
+### Key Idea: **OpenCode = Open-Source AI Coding Agent**
+
+* OpenCode is an **AI coding agent**.
+* It helps create code and files automatically.
+* It is different from a normal AI chatbot.
+
+### Important Term
+
+**AI Coding Agent** → An AI tool that can perform actions like creating code, files, and applications. 
+
+---
+
+# 2. AI Agent vs AI Chat Application
+
+### AI Chat Application
+
+Examples:
+
+* ChatGPT
+* Claude
+* Gemini
+
+What it does:
+
+* Answers questions.
+* Generates text responses.
+* Generates images.
+
+### AI Agent
+
+Examples:
+
+* OpenCode
+* Claude Code
+* Codex
+
+What it does:
+
+* Creates files.
+* Writes code.
+* Uses tools.
+* Performs tasks automatically.
+
+### Important Idea
+
+**Chat App = Gives answers**
+**AI Agent = Does work for you** 
+
+---
+
+# 3. How AI Agents Work
+
+An AI Agent has three major parts:
+
+### 1. LLM (Brain)
+
+* Understands instructions.
+* Makes decisions.
+
+### 2. Tools
+
+Examples:
+
+* Search tools
+* MCP tools
+* Other external tools
+
+### 3. Autonomy
+
+* Works automatically.
+* Continues tasks in a loop.
+* Requires less human intervention.
+
+### Important Formula
+
+**AI Agent = LLM + Tools + Autonomy** 
+
+---
+
+# 4. Why Every AI Agent Needs a Brain
+
+### Key Concept: Large Language Model (LLM)
+
+The transcript repeatedly emphasizes:
+
+> Every AI application or AI agent needs a brain.
+
+The brain is:
+
+* GPT models
+* Claude models
+* Qwen models
+* DeepSeek models
+
+### Important Idea
+
+**LLM = Brain of the AI System** 
+
+---
+
+# 5. Local LLM vs Cloud LLM
+
+There are two ways to connect an AI agent to a brain.
+
+## Local LLM
+
+Runs on your computer.
+
+Benefits:
+
+* More privacy.
+* No cloud dependency.
+
+## Cloud LLM
+
+Runs on provider servers.
+
+Benefits:
+
+* Easier setup.
+* Access to powerful models.
+
+### Important Idea
+
+**AI Agent → Connects to either Local LLM or Cloud LLM** 
+
+---
+
+# 6. Popular Cloud LLM Providers
+
+The transcript mentions:
+
+* Anthropic
+* OpenAI
+* OpenRouter
+* Groq
+
+### Important Idea
+
+These providers supply the **LLM brain** used by AI agents. 
+
+---
+
+# 7. Why OpenRouter is Used
+
+### What OpenRouter Provides
+
+OpenRouter gives access to many AI models from one place.
+
+Examples:
+
+* Claude
+* GPT
+* Qwen
+* DeepSeek
+* GLM
+* Minimax
+
+### Benefit
+
+One API key can provide access to multiple models. 
+
+---
+
+# 8. Installing OpenCode on Windows
+
+### Recommended Setup
+
+Use:
+
+* Windows
+* WSL (Windows Subsystem for Linux)
+* Ubuntu WSL
+* VS Code
+
+### Steps
+
+1. Open Ubuntu WSL.
+2. Create a project folder.
+3. Open the folder in VS Code.
+4. Install OpenCode.
+5. Launch OpenCode.
+
+### Important Idea
+
+The transcript recommends **WSL + Ubuntu** for Windows users. 
+
+---
+
+# 9. Connecting OpenCode to an LLM
+
+After installing OpenCode:
+
+### Step 1
+
+Choose an LLM provider.
+
+Example:
+
+* OpenRouter
+
+### Step 2
+
+Generate an API Key.
+
+### Step 3
+
+Connect OpenCode with the provider.
+
+### Important Idea
+
+**OpenCode (Agent) + LLM (Brain) = Working AI Coding Assistant** 
+
+---
+
+# 10. Models Discussed
+
+## Minimax M3
+
+Features:
+
+* 1 Million context window.
+* Multimodal input.
+* Text output.
+
+---
+
+## Qwen Models
+
+Examples:
+
+* Qwen 3 Max
+* Qwen Code
+
+Features:
+
+* Large context windows.
+* Coding capabilities.
+
+---
+
+## GLM Models
+
+Features:
+
+* Large context size.
+* Available through OpenRouter.
+
+---
+
+## DeepSeek Models
+
+Features:
+
+* Coding-focused models.
+* Available via OpenRouter.
+
+### Important Idea
+
+Different models provide different:
+
+* Cost
+* Speed
+* Context length
+* Coding performance 
+
+---
+
+# 11. Creating an OpenRouter API Key
+
+### Steps
+
+1. Sign up.
+2. Go to API Keys.
+3. Create a new key.
+4. Set limits if required.
+5. Copy the API key.
+
+### Important Idea
+
+The API key allows OpenCode to communicate with the chosen LLM. 
+
+---
+
+# 12. Plan Mode vs Build Mode
+
+## Plan Mode
+
+* Gives explanations.
+* Produces text output.
+* Does not create files.
+
+## Build Mode
+
+* Creates files.
+* Writes code.
+* Modifies the repository.
+
+### Important Idea
+
+**Plan = Think**
+**Build = Create** 
+
+---
+
+# 13. Practical Demo
+
+Prompt used:
+
+> Create a Dockerfile for a Python app in the current repository.
+
+OpenCode:
+
+* Analyzed the repository.
+* Asked clarifying questions.
+* Generated a Dockerfile.
+* Saved it in the project.
+
+### Important Idea
+
+OpenCode can generate real project files automatically. 
+
+---
+
+# 14. Switching Models
+
+OpenCode allows changing models.
+
+Examples mentioned:
+
+* DeepSeek
+* Neotron
+* Qwen
+* Claude
+* GPT
+
+### Benefit
+
+You can choose:
+
+* Free models
+* Paid models
+* Faster models
+* More capable models
+
+
+
+---
+
+# 15. Free Models
+
+The transcript highlights:
+
+* Free models are available through OpenRouter.
+* Example: Nvidia Neotron models.
+* Free models may respond slower than paid models.
+
+### Important Idea
+
+**Free models save cost but may be slower.** 
+
+---
+
+# Final Beginner Takeaway
+
+### Remember These 5 Points
+
+1. **OpenCode is an open-source AI coding agent.**
+2. **AI Agents perform actions; chatbots mainly provide responses.**
+3. **Every AI agent needs an LLM brain.**
+4. **OpenRouter lets you connect many LLMs using one API key.**
+5. **OpenCode can create real files (like Dockerfiles) directly in your project.**
+
+### Simple Formula
+
+**OpenCode (Agent) + OpenRouter (Provider) + LLM (Brain) = AI-Powered Coding Assistant**