🌸

Unnati Tripathi

🌸 Updated May 2026

Your AI Survival Guide

Everything you need to learn about AI tools, agents, and workflows — whether you're a student, marketer, designer, founder, or someone who just wants to keep their job.

Start Here

What Do These Words Even Mean?

.md Files

Markdown files — plain text documents that give AI structured instructions. Think of them as your AI's cheat sheet.

Obsidian

A note-taking app that stores everything as .md files. Used to build a personal knowledge base that feeds into AI workflows.

Skill Files

Pre-written instruction files (SKILL.md) that tell AI agents exactly how to complete specific tasks — like a job description for AI.

Context Window

Everything an AI can "see" at once — your prompt, files, history. Bigger context = smarter, more aware responses.

MCP

Model Context Protocol — the "USB-C of AI." A universal standard that lets any AI model connect to any tool, API, or data source.

Connectors

Bridges between AI and your real apps — Gmail, Slack, Notion, Calendar. AI agents use connectors to take real actions in your world.

APIs

Application Programming Interfaces — how apps talk to each other. Every AI tool runs on APIs. Learning APIs = learning the internet's backbone.

RAG

Retrieval-Augmented Generation — AI that searches real data before answering. No hallucinations. Like giving AI access to Google before it speaks.

Vector Databases

Storage systems that hold data as math (vectors). AI searches by meaning, not keywords. Powers RAG, memory, and semantic search.

LangChain

The most popular framework for building AI apps. Chains together LLMs, tools, and memory into complete workflows and agents.

AI Agents

AI that doesn't just answer — it acts. Give it a goal, it plans, executes, and completes tasks autonomously without you babysitting it.

Multi-Agent Orchestration

A boss agent that manages other agents. One instruction triggers an entire AI workforce. This is replacing middle management.

OpenClaw

Viral AI agent (200K+ GitHub stars in weeks) that lives in WhatsApp, Telegram & Slack and actually does things — books, emails, manages files.

Claude Code

Anthropic's terminal-based coding agent. Reads your codebase, writes code, runs tests, and opens pull requests — all from the command line.

Cursor

The most hyped AI code editor of 2026. Built on VS Code, it understands your entire codebase — not just one file — and edits across multiple files at once.

Vibe Coding

Building software by describing what you want in plain English. AI writes the code. You just vibe. No syntax, no Stack Overflow. Just ship.

Your Path

Learn This in 90 Days 🗓️

Week 1–2

Understand What AI Actually Is

Start with how LLMs work, what prompting means, and why context matters. No code needed. Just conceptual understanding. Use ChatGPT or Claude daily.

Week 3–4

Set Up Obsidian + .md Files

Build your second brain. Learn Markdown syntax. Start writing .md files for your notes, ideas, and workflows. This becomes your AI knowledge base.

Week 5–6

Learn What APIs & MCP Do

Understand how apps connect. Try connecting Claude to a tool via MCP (no-code options available). See connectors in action with Gmail or Notion.

Week 7–8

Explore RAG & Vector Databases

Take a free RAG course. Understand how AI retrieves data. Try building a simple Q&A bot on your own documents using free tools.

Week 9–10

Try Vibe Coding with Cursor or Claude Code

Build something. A landing page, a tool, an app. Describe it in English and let AI build it. No prior coding needed. Just try Bolt.new or Lovable first.

Week 11–12

Build Your First AI Agent

Use LangChain or n8n to build an agent that automates one task in your life. Email summarizer, meeting scheduler, content drafter. Start small, go big.

Technical

How It All Connects (With Real Code)

RAG Pipeline

How RAG Actually Works

RAG = Retrieval + Generation. The AI searches a vector database for relevant chunks of your data, injects them into the context window, then generates an answer. No hallucinations.

# Simple RAG pipeline with LangChain
from langchain.vectorstores import Chroma
from langchain.embeddings import OpenAIEmbeddings
from langchain.chains import RetrievalQA

# 1. Store your docs as vectors
vectorstore = Chroma.from_documents(docs, OpenAIEmbeddings())

# 2. Build a retriever
retriever = vectorstore.as_retriever()

# 3. Ask questions against YOUR data
qa = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)
answer = qa.run("What does my document say about X?")
      

MCP

How MCP Connects AI to Tools

MCP uses a client-server architecture. Your AI model is the client. Every tool (Gmail, Notion, GitHub) is a server. MCP is the protocol they speak to each other.

# MCP server in Python (exposes a tool to any AI)
from mcp import FastMCP

mcp = FastMCP("My Tool Server")

@mcp.tool()
def send_email(to: str, subject: str, body: str) -> str:
    """Send an email via AI agent"""
    # Your email logic here
    return f"Email sent to {to}"

# Now ANY AI model with MCP support can send emails
mcp.run()
      

AI Agents

Building a Simple AI Agent

An agent has a goal, a set of tools, and a loop: Think → Act → Observe → Repeat. It keeps going until the task is done. No human needed after the first prompt.

# AI Agent with LangChain
from langchain.agents import initialize_agent, Tool
from langchain.tools import DuckDuckGoSearchRun

search = DuckDuckGoSearchRun()
tools = [Tool(
    name="Search",
    func=search.run,
    description="Search the web for info"
)]

agent = initialize_agent(
    tools=tools, llm=llm,
    agent="zero-shot-react-description"
)

# Agent will plan & execute autonomously
agent.run("Research the top AI tools in 2026 and summarise")
      

Vibe Coding

How Vibe Coding Works

You describe what you want in plain English. Claude Code or Cursor reads your codebase, plans the changes, writes the code, and executes. You review and approve.

# In your terminal with Claude Code:
$ claude

# Then just describe what you want:
"Build me a REST API with FastAPI that has
 user auth, a /dashboard endpoint, and
 connects to a PostgreSQL database.
 Write tests and deploy to Railway."

# Claude Code will:
# → Read your existing code
# → Plan the implementation
# → Write all the files
# → Run the tests
# → Fix errors automatically
      