Code-Examples

These are code examples of langgraph presentation

import userdata from google.colab import requests

ENV VARIABLES for security

SERPER_API_KEY = userdata.get("SERPER_API_KEY") GEMINI_API_KEY = userdata.get("GEMINI_API_KEY") # Hypothetical for Gemini TAVILY_API_KEY = userdata.get("TAVILY_API_KEY") # Hypothetical Tavily

#####################

Tavily Node

##################### def tavily_transform(input_text): """ Hypothetical function using Tavily's library to transform/clean/format text. """ # ... implement your Tavily logic here ... return f"[Tavily Pre-Processed]\n{input_text}"

#####################

Serper Search Node

##################### def serper_search(query): url = "https://google.serper.dev/search" headers = { "X-API-KEY": SERPER_API_KEY, "Content-Type": "application/json" } payload = {"q": query}

response = requests.post(url, headers=headers, json=payload)
data = response.json()
# Return top snippet or entire result
if "organic" in data and len(data["organic"]) > 0:
    return data["organic"][0].get("snippet", "")
return "No search results found."

#####################

Gemini Node (LLM)

##################### def gemini_generate(prompt): """ Pseudocode for calling Gemini's API. Replace with the actual method once Gemini's public API is available. """ # Typically a POST request with your GEMINI_API_KEY # For demonstration, returning a mock response: return f"[Gemini Output]: Response to your prompt -> {prompt}"

#####################

Human in the Loop (HITL)

##################### def human_review(ai_output): """ In a real system, this might pause and allow human intervention. For now, we simulate auto-approval. """ print("HITL: Checking AI output ...") # e.g., a user interface prompt or form approval = True # or False if the human disapproves return ai_output if approval else "[Human Overrode Output]"

#####################

Orchestrate the Graph

##################### def language_graph_pipeline(user_input): # 1. Tavily Transform tavily_output = tavily_transform(user_input)

# 2. Serper Search (Optional) - if you want external context
search_output = serper_search(user_input)

# 3. Combine everything into a final prompt
final_prompt = f"{tavily_output}\nSearch Context: {search_output}"

# 4. Gemini Generation
gemini_response = gemini_generate(final_prompt)

# 5. Human Review
final_answer = human_review(gemini_response)

return final_answer

Example usage:

if name == "main": user_query = "What are the best AI chatbot frameworks in 2025?" result = language_graph_pipeline(user_query) print("\n=== Final Result ===") print(result)