Building AI Agents with Python: A Deep Dive

This covers Ai agents from a builder's perspective. It covers:

- Building agents from scratch
- Agentic concepts
- Building agents
- Agents working together

Transcript

1. Building Ai agents in Python: A Deep Dive
3. About me • Abdur-Rahmaan Janhangeer • Been working with Python professionally since a long time • Worked for companies in all continents (except Antartica and South America) • Pymug organizing member compileralchemy.com
4. Slides
5. Building Ai agents in Python: A Deep Dive
6. Building agents from real scratch
7. Task: Control a drone with your mind
8. Task: Control a drone with your mind
9. Task: Control a drone with your mind Use openai or Gemini API to take off or land a drone
10. Task: Control a drone with your mind Use openai or Gemini API to take off or land a drone
11. Task: Control a drone with your mind Use openai or Gemini API to take off or land a drone
12. Task: Control a drone with your mind Use openai or Gemini API to take off or land a drone
13. Task: Control a drone with your mind Function Calling (old) def get_headset_value(): … def move_drone_up(): … def move_drone_down(): …
14. Task: Control a drone with your mind import openai import time import random def get_brainwave_value(): """Simulate getting a brainwave focus value from the NeuroSky headset.""" return random.randint(0, 100) def move_drone_up(): print("Drone moving up") def move_drone_down(): print("Drone moving down")
15. Task: Control a drone with your mind functions = [ { "name": "get_brainwave_value", "description": "Get the current brainwave focus value.", "parameters": {} }, { "name": "move_drone_up", "description": "Move the drone up on high focus.", "parameters": {} }, { "name": "move_drone_down", "description": "Move the drone down on low focus.", "parameters": {} } ]
16. Task: Control a drone with your mind def control_drone(): """Continuously monitor brainwave focus values and move the drone accordingly.""" while True: value = get_brainwave_value() print(f"Brainwave focus value: {value}") response = openai.ChatCompletion.create( model="gpt-4-turbo", messages=[ {"role": "system", "content": "You control a drone based on brainwave focus levels."}, {"role": "user", "content": value} ], functions=functions, # here function_call="auto" ) function_name = response["choices"][0]["message"].get("function_call", {}).get("name")
17. Task: Control a drone with your mind if function_name == "move_drone_up": move_drone_up() elif function_name == "move_drone_down": move_drone_down() time.sleep(1) if __name__ == "__main__": control_drone() Even the base version has enough meat
18. Raw function calling is powerful
19. Task: Create a chatGPT plugin system
20. Task: Create a chatGPT plugin system
21. Task: Create a chatGPT plugin system
22. Task: Create a chatGPT plugin system
23. Task: Create a chatGPT plugin system
24. Task: Create a chatGPT plugin system
25. Raw function calling is powerful
26. Raw function calling is powerful Even when dumb
27. The concept of agents
28. Function calling is the meat of Ai agents
29. Function calling is the meat of Ai agents
30. Overview of an agent
31. ReAct
32. Generally speaking
33. Flows
34. Chaining of actions
35. Gating
36. Process results
37. Feedback
38. Building agents
39. Smolagent
40. Smolagent from typing import Optional import requests from smolagents import CodeAgent, LiteLLMModel, tool model = LiteLLMModel(model_id="gpt-4o")
41. Smolagent @tool def get_joke() -> str: """ Fetches a random joke from the JokeAPI. This function sends a GET request to the JokeAPI to retrieve a random joke. It handles both single jokes and two-part jokes (setup and delivery). If the request fails or the response does not contain a joke, an error message is returned. Returns: str: The joke as a string, or an error message if the joke could not be fetched. """ url = "https://v2.jokeapi.dev/joke/Any?type=single"
42. Smolagent try: response = requests.get(url) response.raise_for_status() data = response.json() if "joke" in data: return data["joke"] elif "setup" in data and "delivery" in data: return f"{data['setup']} - {data['delivery']}" else: return "Error: Unable to fetch joke." except requests.exceptions.RequestException as e: return f"Error fetching joke: {str(e)}"
43. Smolagent agent = CodeAgent( tools=[ get_joke, ], model=model, ) agent.run("Tell me a joke")
44. Agents working together
45. Agents for tasks
46. Let’s say we need to find the phone numbers of hardware shops
47. Let’s say we need to find the phone numbers of hardware shops
48. Let’s say we need to find the phone numbers of hardware shops
49. Let’s say we need to find the phone numbers of hardware shops
50. Let’s say we need to find the phone numbers of hardware shops
51. Let’s say we need to find the phone numbers of hardware shops
52. Graph Agents
53. Graph Agents - Using langgraph - https://langchain-ai.github.io/langgraph/ - State - Node
54. Graph Agents State as in state machine
55. Graph Agents State as in state machine
56. Graph Agents State as in state machine
57. Graph Agents State as in state machine, but with a twist
58. Graph Agents Unit of work are called nodes
59. Graph Agents Unit of work are called nodes Nodes can have edges The indicate the control flow and order of execution
60. Graph Agents Unit of work are called nodes Nodes can have edges Conditional edges indicate branching
61. Graph Agents Unit of work are called nodes They can be agents
62. Graph Agents Unit of work are called nodes Nodes modify state/s
63. Graph Agents Unit of work are called nodes Nodes modify state/s
64. Graph Agents Unit of work are called nodes Nodes modify state/s The state is the input to any node whenever it is invoked
65. Graph Agents Unit of work are called nodes Nodes modify state/s The state is the input to any node whenever it is invoked
66. Graph Agents Unit of work are called nodes Nodes modify state/s The state is the input to any node whenever it is invoked
67. Graph Agents Unit of work are called nodes Nodes modify state/s The state is the input to any node whenever it is invoked
68. Graph Agents Unit of work are called nodes Nodes modify state/s The state is the input to any node whenever it is invoked
69. Graph Agents Unit of work are called nodes Nodes modify state/s The state is the input to any node whenever it is invoked
70. Graph Agents langgraph is powerful enough It can be used to build pretty much any agent scenario or reasoning-focused workflows
71. Graph Agents langgraph is powerful enough It can be used to build pretty much any agent scenario or reasoning-focused workflows BUT SINCE ITS FROM LANGCHAIN, THE API IS AS BAD AS IT CAN BE
72. MCP
73. What if we could make agents talk to each other?
74. What if there was a standardised approach to making agents talk to each other?
75. That’s where MCP comes in
76. Model Context Protocol
77. Model Context Protocol
78. Model Context Protocol
79. Model Context Protocol
80. Model Context Protocol
81. Model Context Protocol
82. Through JSON-RPC
83. Python $ pip install mcp
84. Python # server.py from mcp.server.fastmcp import FastMCP # Create an MCP server mcp = FastMCP("Demo") # Add an addition tool @mcp.tool() def add(a: int, b: int) -> int: """Add two numbers""" return a + b # Add a dynamic greeting resource @mcp.resource("greeting://{name}") def get_greeting(name: str) -> str: """Get a personalized greeting""" return f"Hello, {name}!"
85. Python (Claude desktop) $ mcp install server.py
86. MCP: A Game Changer
87. Marketplace? Just like people sell APIs on marketplaces
88. Enabling agents Search registry for tools
89. Enabling agents Search registry for tools
90. Watch out It’s the web so …
91. Watch out It’s the web so … Think about security for example
92. Papers
93. Papers - ReAct: ReAct: Synergizing Reasoning and Acting in Language Models - ReWOO: ReWOO: Decoupling Reasoning from Observations for Efficient Augmented Language Models - LLM Compiler: An LLM Compiler for Parallel Function Calling
94. Interesting reads Abhinav Upaday - chatGPT plugin Anthropic - Effective agents https://github.com/mannaandpoem/OpenManus See how to use Python with n8n
95. That’s all folks! Join the Pymug on WA