Lesson 07 � Advanced Skill
AI AGENTS:
KHUD SE DECISIONS LENE WALE AI.
AI agents woh hain jo khud se sochte hain, tools use karte hain, aur tasks complete karte hain. Jaise ek real assistant � aap bolo kaam kya hai, woh khud figure out karega kaise karna hai. Yeh AI ka future hai.
WHY: AI Agents kyun zaroori hain?
Normal AI models sirf text generate karte hain. Lekin agents karte hain � yeh sochte hain, tools use karte hain, aur tasks complete karte hain. Imagine karo: aapne bola "mera data analyze karo aur report banao" � agent khud se code likhega, data load karega, analyze karega, aur report generate karega. No human intervention needed!
External capabilities jo agent use kar sakta hai � web search, code execution, API calls, file operations. Agent ke haath-pair hain tools.
Task decomposition � bade task ko chhote steps mein todna. Agent sochta hai ki pehle kya karna hai, phir kya, phir kya.
Actions lene ka step. Planning ke baad agent actually tools call karta hai aur kaam karta hai.
Context retention � agent ko yaad rehta hai ki pehle kya kiya, kya results aaye. Isse woh better decisions leta hai.
Agent ka Basic Concept
Ek agent ke paas hota hai naam, tools ki list, aur memory. Woh sochta hai (think), act karta hai (act), aur results store karta hai. Simple hai � jaise ek human worker!
# Simple agent concept
class Agent:
def __init__(self, name, tools):
self.name = name
self.tools = tools
self.memory = []
def think(self, task):
print(f"{self.name} is thinking about: {task}")
return f"Plan for: {task}"
def act(self, tool_name, input_data):
if tool_name in self.tools:
print(f"{self.name} using {tool_name}")
return f"Result from {tool_name}"
return "Tool not found"
# Create agent
agent = Agent("DSWallah Bot", ["search", "calculate", "write"])
print(agent.think("Answer student query"))
print(agent.act("search", "Python tutorial"))
# Real agent with tools
tools = ["web_search", "code_execute", "file_read", "api_call"]
agent = Agent("Data Scientist", tools)
print(f"Tools available: {agent.tools}")Agents Kaise Kaam Karte Hain?
Agent ka kaam hota hai: observe ✓ think ✓ act ✓ repeat. Yeh cycle tab tak chalti hai jab tak task complete na ho jaye.
Agent environment ko dekhta hai � kya data hai, kya tools available hain, kya task karna hai.
Agent plan banata hai � kaunsa tool use karna hai, pehle kya karna hai, phir kya.
Agent tool call karta hai aur kaam karta hai. Result milta hai.
Agar task complete nahi hua, toh woh phir se observe ✓ think ✓ act karta hai. Tab tak jab tak kaam na ho jaye.
Real Agent Example
Yeh ek real-world agent hai jo data science tasks handle karta hai. Tools use karta hai aur memory mein results store karta hai.
# Real agent with multiple tools and memory
class RealAgent:
def __init__(self, name):
self.name = name
self.tools = {
"search": self.search_tool,
"calculate": self.calculate_tool,
"analyze": self.analyze_tool
}
self.memory = []
def search_tool(self, query):
return f"Found results for: {query}"
def calculate_tool(self, expression):
try:
result = eval(expression)
return f"Calculation result: {result}"
except:
return "Invalid expression"
def analyze_tool(self, data):
return f"Analysis complete for {len(data)} items"
def think(self, task):
plan = f"Plan: Break '{task}' into steps"
self.memory.append({"task": task, "plan": plan})
return plan
def act(self, tool_name, input_data):
if tool_name in self.tools:
result = self.tools[tool_name](input_data)
self.memory.append({"tool": tool_name, "result": result})
return result
return f"Tool '{tool_name}' not found"
def show_memory(self):
print(f"\n{self.name}'s Memory:")
for item in self.memory:
print(f" - {item}")
# Create and use agent
ds_agent = RealAgent("DataScientist")
print(ds_agent.think("Analyze sales data"))
print(ds_agent.act("search", "sales trends 2024"))
print(ds_agent.act("calculate", "1000 * 1.15"))
print(ds_agent.act("analyze", [1, 2, 3, 4, 5]))
ds_agent.show_memory()Try it: Build Your Agent
Yahan apna agent banao aur dekho woh kaise kaam karta hai. Tools add karo, tasks do, aur dekho agent kaise sochta hai aur act karta hai.
Exercise: Test Your Knowledge
Quick check
"AI Agent kya hai?"
Socho: normal AI aur agent mein kya fark hai✓ Agent kya special karta hai?
Key Takeaways
- Agents autonomous hain: Yeh khud se decisions lete hain, human ki har step par zaroorat nahi.
- Tools are everything: Agent ka power uske tools mein hai � search, calculate, code, API calls.
- Planning is key: Bina plan ke agent confused ho jayega. Task ko steps mein todna zaroori hai.
- Memory matters: Agent ko yaad rehta hai ki pehle kya kiya � isse woh better decisions leta hai.
- Loop continues: Agent tab tak kaam karta hai jab tak task complete na ho jaye � observe ✓ think ✓ act ✓ repeat.
Ab ethics par chalo � taaki aap samajh sako ki AI agents ka responsible use kaise karna hai aur kya risks hain.