Project Ideas :

Create a code reviewer like Code Rabbit, which has a UI to paste a link and should review PR/MR from GitHub or Gitlab
Mock Interviewer > Take 2 inputs: job requirements/role & responsibilities, and Resume. AI asks questions to you using a voice model - Feedback on answers and rating
Article Reviewer - Paste your hashcode article link - AI finds out mistakes in it, if any - Suggest feedback and rate the articles

There are a lot of problems to be solved by AI.
Value-driven products are what we need. Real-world problems. NLP to NLP.
Solve a problem for yourself.
How to evaluate a new idea? Find Delta and how much value does it add?
Delta = Value provided by your solution, system, or process - Value provided by your solution, system, or process
State A: 2 and State B: 8. Delta = State B - State A | Delta >= 4 for creating massive wealth

Advantages of Delta 4 behavior:

Creates an irreversible effect - The User would not go back to the old behavior
UBP > USP: Barging behavior - Mouth-to-mouth-free publicity, so we need less fuel for acquiring new users
High Tolerance: The User would tolerate your mistakes only if the delta is greater than 4

Best Explanation for Delta: https://youtu.be/fWm_NKCtXdA?si=wjKdPju31hZ9vuKf
The higher the Delta, the better the product or service you're building
Humans consistently seek efficiency. Remove the hurdle and build wealth as an entrepreneur

Bluelearn - an AI agent company that solves problems using day-to-day tools like WhatsApp

Provides a framework for your code
Langchain is a library, but LangGraph is a framework
For example, FastAPI - when you add @app.post above your function, then it runs as LangGraph works. When we use the library, we decide the flow on our own, using imports and tools we built
LangGraph provides a structure for your code
Due to libraries, scaling code for an AI application becomes difficult to maintain. That’s why we use LangGraph to make code more modular.
The developer decides the flow.
LangGraph suggests that we change the structure of our code into a graph, so adding and removing components from the code becomes easy, and the code is structured in a modular format
Write code and connect it using nodes, which in turn become a flow
n8n is kind of the UI of LangGraph

LangGraph orchestration :

To the first node of LangGraph, we provide a state (data) to it.
Updating the state after each node
Set up tracing to truly understand how the routing works and how the graph is formed.
pydentic adds a new validation layer. It’s kind of typescipt for Python code.
Plan > Think > Observe > Output

Using LangGraph, we need to create states and nodes, and then we need to update the states based on the flow of the application.

Assignment: Build the flow for the cursor project

LangGraph - Orchestration Framework