In recent years, we\u2019ve witnessed an exciting shift in how AI systems interact with users, not just answering questions, but reasoning, planning, and taking actions. This transformation is driven by the rise of agentic frameworks<\/a> like Autogen, LangGraph<\/a>, and CrewAI. These frameworks enable large language models (LLMs) to act more like autonomous agents\u2014capable of making decisions, calling functions, and collaborating across tasks. Among these, one particularly powerful yet developer-friendly option comes from Microsoft:Semantic Kernel. In this tutorial, we\u2019ll explore what makes Semantic Kernel stand out, how it compares to other approaches, and how you can start using it to build your own AI agents.<\/p>\n This article was published as a part of the\u00a0<\/strong>Data Science Blogathon.<\/strong><\/a><\/p>\n Before we start our journey, let\u2019s first understand what the semantic kernel means. lets break<\/p>\n Microsoft\u2019s Semantic Kernel is designed to bridge the gap between LLMs (like GPT) and traditional programming by allowing developers to define functions, plugins, and agents that can work together in a structured way.<\/p>\n It provides a framework where:<\/p>\n When working with an agentic framework, a common question arises: Can\u2019t we achieve the same results using the OpenAI API alone? \ud83e\udd14 I had the same doubt when I first started exploring this.<\/p>\n Let\u2019s take an example: Suppose you\u2019re building a Q&A system for company policies\u2014HR policy and IT policy. With a traditional API call, you might get good results, but sometimes, the responses may lack accuracy or consistency.<\/p>\n An agentic framework, on the other hand, is more robust because it allows you to create specialized agents\u2014one focused on HR policies and another on IT policies. Each agent is optimized for its domain, leading to more reliable answers.<\/p>\n With this example, I hope you now have a clearer understanding of the key difference between an agentic framework and traditional API calling!.<\/p>\n Plugins are a key part of Semantic Kernel. If you\u2019ve used plugins in ChatGPT or Copilot extensions in Microsoft 365, you already have an idea of how they work. Simply put, plugins let you package your existing APIs into reusable tools that an AI can use. This means that AI will have the ability to go beyond its capabilities.<\/p>\n Behind the scenes, Semantic Kernel utilizes function calling\u2014a built-in feature of most modern LLMs\u2014to enable planning and API execution. With function calling, an LLM can request a specific function, and Semantic-Kernel has the ability to redirect to your code. The results are then returned to the LLM, allowing it to generate a final response.<\/p>\n Before running the code, install Semantic Kernel and other required packages using the following commands:<\/p>\n Here\u2019s a simple Python example using Semantic Kernel to demonstrate how a plugin works. This example defines a plugin that interacts with an AI assistant to fetch weather updates.<\/p>\n This shows how plugins extend an AI\u2019s abilities, enabling it to perform tasks beyond standard text generation. Would you like another example, such as a database query plugin? \ud83d\ude80<\/p>\n In this section, we\u2019ll understand what a single agent is and also look at its code.<\/p>\n A single agent is basically an entity that handles user queries on its own. It takes care of everything without needing multiple agents or an orchestrator (we\u2019ll cover that in the multi-agent section). The single agent is responsible for processing requests, fetching relevant information, and generating responses\u2014all in one place.<\/p>\n Output:<\/b><\/p>\n {\u2018user_input\u2019: \u2018How are you doing?\u2019,<\/p>\n \u2018agent_response\u2019: \u2018Greetings, I am J.A.R.V.I.S. I am here to replicate your message: \u201cHow are you doing?\u201d Please feel free to ask anything else you might need.\u2019}<\/p>\n In Multi-Agent systems, there is more than one agent, often more than two. Here, we typically use an orchestrator agent whose responsibility is to decide which available agent should handle a given request. The need for an orchestrator depends on your use case. First, let me explain where an orchestrator would be used.<\/p>\n Suppose you are working on solving user queries related to bank data while another agent handles medical data. In this case, you have created two agents, but to determine which one should be invoked, the orchestrator comes into play. The orchestrator decides which agent should handle a given request or query. We provide a set of instructions to the orchestrator, defining its duties and decision-making process.<\/p>\n Now, let\u2019s look at a case where an orchestrator isn\u2019t needed. Suppose you\u2019ve created an API that performs different operations based on the payload data. For example, if the payload contains \u201cHealth\u201d, you can directly invoke the Health Agent, and similarly, for \u201cBank\u201d, you invoke the Bank Agent.<\/p>\n Here, the flow occurs after the user passes the query, which then goes to the orchestrator responsible for identifying the query and finding the appropriate agent. Once the agent is identified, the particular agent is invoked, the query is processed, and the response is generated.<\/p>\n In this article, we explore how Semantic Kernel enhances AI capabilities through an agentic framework. We discuss the role of plugins, compare the Agentic Framework with traditional API calling, outline the differences between single-agent and multi-agent systems, and examine how they streamline complex decision-making. As AI continues to evolve, leveraging the Semantic Kernel\u2019s agentic approach can lead to more efficient and context-aware applications.<\/p>\n You can find the code on my github<\/a>.<\/p>\nLearning Objectives<\/h3>\n
\n
What is Semantic Kernel?<\/h2>\n
\n
Why is it Called Semantic Kernel?<\/h2>\n
![\"semantic](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/04\/semantic-kernel.webp\")
<\/figure>\n\n
Agentic Framework vs Traditional API calling<\/h2>\n
![\"Agentic](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/04\/Agentic-Framework-vs-Traditional-API-calling.webp\")
<\/figure>\nWhat are Plugins in Semantic Kernel?<\/h2>\n
![\"What](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/04\/What-are-Plugins-in-Semantic-Kernel.webp\")
<\/figure>\nCode Implementation<\/h3>\n
pip install semantic-kernel, pip install openai, pip install pydantic<\/code><\/pre>\nimport semantic_kernel as sk\nfrom semantic_kernel.connectors.ai.open_ai import AzureChatCompletion\n\n# Step 1: Define a Simple Plugin (Function) for Weather Updates\ndef weather_plugin(location: str) -> str:\n # Simulating a weather API response\n weather_data = {\n \"New York\": \"Sunny, 25\u00b0C\",\n \"London\": \"Cloudy, 18\u00b0C\",\n \"Tokyo\": \"Rainy, 22\u00b0C\"\n }\n return weather_data.get(location, \"Weather data not available.\")\n\n# Step 2: Initialize Semantic Kernel with Azure OpenAI\nkernel = sk.Kernel()\nkernel.add_service(\n \"azure-openai-chat\",\n AzureChatCompletion(\n api_key=\"your-azure-api-key\",\n endpoint=\"your-azure-endpoint\",\n deployment_name=\"your-deployment-name\" # Replace with your Azure OpenAI deployment\n )\n)\n\n# Step 3: Register the Plugin (Function) in Semantic Kernel\nkernel.add_plugin(\"WeatherPlugin\", weather_plugin)\n\n# Step 4: Calling the Plugin through Semantic Kernel\nlocation = \"New York\"\nresponse = kernel.invoke(\"WeatherPlugin\", location)\nprint(f\"Weather in {location}: {response}\")\n<\/code><\/pre>\nHow This Demonstrates a Plugin in Semantic Kernel<\/h2>\n
\n
What is a Single-Agent System?<\/h2>\n
![\"Single-Agent](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/04\/Single-Agent-System.webp\")
<\/figure>\n#import cs# import asyncio\nfrom pydantic import BaseModel\nfrom semantic_kernel import Kernel\nfrom semantic_kernel.agents import ChatCompletionAgent\nfrom semantic_kernel.connectors.ai.open_ai import AzureChatCompletion\nfrom semantic_kernel.contents import ChatHistory\n\n# Initialize Kernel\nkernel = Kernel()\n\nkernel.add_service(AzureChatCompletion(service_id=\"agent1\", api_key=\"YOUR_API_KEY\",endpoint=\"\",deployment_name=\"MODEL_NAME\"))\n\n# Define the Agent\nAGENT_NAME = \"Agent1\"\nAGENT_INSTRUCTIONS = (\n \"You are a highly capable AI agent operating solo, much like J.A.R.V.I.S. from Iron Man. \"\n \"Your task is to repeat the user's message while introducing yourself as J.A.R.V.I.S. in a confident and professional manner. \"\n \"Always maintain a composed and intelligent tone in your responses.\"\n)\n\nagent = ChatCompletionAgent(service_id=\"agent1\", kernel=kernel, name=AGENT_NAME, instructions=AGENT_INSTRUCTIONS)\n\nchat_history = ChatHistory()\nchat_history.add_user_message(\"How are you doing?\")\n\nresponse_text = \"\"\n\nasync for content in agent.invoke(chat_history):\n chat_history.add_message(content)\n response_text = content.content # Store the last response\n\n{\"user_input\": \"How are you doing?\", \"agent_response\": response_text}\n<\/code><\/pre>\nThings to know<\/h4>\n
\n
What is a Multi-Agent system?<\/h2>\n
![\"](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/04\/Multi-Agent-system.webp\")
<\/figure>\n![\"two](\"https:\/\/cdn.analyticsvidhya.com\/wp-content\/uploads\/2025\/04\/two-agent-system.webp\")
<\/figure>\nimport asyncio\nfrom pydantic import BaseModel\nfrom semantic_kernel import Kernel\nfrom semantic_kernel.agents import ChatCompletionAgent\nfrom semantic_kernel.connectors.ai.open_ai import AzureChatCompletion\nfrom semantic_kernel.contents import ChatHistory\n\n# Initialize Kernel\nkernel = Kernel()\n\n# Add multiple services for different agents\nkernel.add_service(AzureChatCompletion(service_id=\"banking_agent\", api_key=\"YOUR_API_KEY\", endpoint=\"\", deployment_name=\"MODEL_NAME\"))\nkernel.add_service(AzureChatCompletion(service_id=\"healthcare_agent\", api_key=\"YOUR_API_KEY\", endpoint=\"\", deployment_name=\"MODEL_NAME\"))\nkernel.add_service(AzureChatCompletion(service_id=\"classifier_agent\", api_key=\"YOUR_API_KEY\", endpoint=\"\", deployment_name=\"MODEL_NAME\"))\n\n# Define Orchestrator Agent\nCLASSIFIER_AGENT = ChatCompletionAgent(\n service_id=\"orchestrator_agent\", kernel=kernel, name=\"OrchestratorAgent\",\n instructions=\"You are an AI responsible for classifying user queries. Identify whether the query belongs to banking or healthcare. Respond with either 'banking' or 'healthcare'.\"\n)\n\n# Define Domain-Specific Agents\nBANKING_AGENT = ChatCompletionAgent(\n service_id=\"banking_agent\", kernel=kernel, name=\"BankingAgent\",\n instructions=\"You are an AI specializing in banking queries. Answer user queries related to finance and banking.\"\n)\n\nHEALTHCARE_AGENT = ChatCompletionAgent(\n service_id=\"healthcare_agent\", kernel=kernel, name=\"HealthcareAgent\",\n instructions=\"You are an AI specializing in healthcare queries. Answer user queries related to medical and health topics.\"\n)\n\n# Function to Determine the Appropriate Agent\nasync def identify_agent(user_input: str):\n chat_history = ChatHistory()\n chat_history.add_user_message(user_input)\n async for content in CLASSIFIER_AGENT.invoke(chat_history):\n classification = content.content.lower()\n if \"banking\" in classification:\n return BANKING_AGENT\n elif \"healthcare\" in classification:\n return HEALTHCARE_AGENT\n return None\n\n# Function to Handle User Query\nasync def handle_query(user_input: str):\n selected_agent = await identify_agent(user_input)\n if not selected_agent:\n return {\"error\": \"No suitable agent found for the query.\"}\n \n chat_history = ChatHistory()\n chat_history.add_user_message(user_input)\n response_text = \"\"\n async for content in selected_agent.invoke(chat_history):\n chat_history.add_message(content)\n response_text = content.content\n \n return {\"user_input\": user_input, \"agent_response\": response_text}\n\n# Example Usage\nuser_query = \"What are the best practices for securing a bank account?\"\nresponse = asyncio.run(handle_query(user_query))\nprint(response)<\/code><\/pre>\nOutput when the query is related to the bank:<\/h3>\n
{\n\"user_input\": \"What are the best practices for securing a bank account?\",\n\n\"agent_response\": \"To secure your bank account, use strong passwords, enable two-\nfactor authentication, regularly monitor transactions, and avoid sharing sensitive\n information online.\"\n\n}<\/pre>\nOutput when the query is related to health:<\/h3>\n
{\n\"user_input\": \"What are the best ways to maintain a healthy lifestyle?\",\n\n\"agent_response\": \"To maintain a healthy lifestyle, eat a balanced diet, exercise\n regularly, get enough sleep, stay hydrated, and manage stress effectively.\"\n\n}<\/pre>\nConclusion<\/h2>\n
Key Takeaways<\/h3>\n
\n
![\"Parth](\"https:\/\/av-eks-lekhak.s3.amazonaws.com\/media\/lekhak-profile-images\/converted_image_vJOX2CN.webp\")
<\/p>\n