In this repository we'll show you how to leverage Semantic Kernel into an existing application. As an example we'll use how you can benefit from Semantic Kernel in a Hotel booking website. You can use natural language to search through a list of hotels.
This demostration was shown on following events:
- [Austria] Oct/23 - Microsoft Build 2023
- [Ireland] Oct/23 - Microsoft Build 2023
- [Belgium] Nov/23 - SaaS local day
- [Austria] Dec/23 - Unlocking AI Opportunities:
Semantic Kernel allows developers to use AI without any knowledge about AI or LLM. As a web developer or software engineer you can focus on the things you're best: writing code for business applications, and you have the LLM or other models completely abstracted.
This example has two parts:
-
an
uipart which contains the Hotel booking website written in React.yarn start
-
an
apipart which contains the Hotel booking API written in Python and contains some custom logic and the Semantic Kernel SDK.python main
Within the api we are using a LLM model, either hosted on OpenAI or Azure OpenAI. Before you can run the backend (as mentioned above), make sure you have defined you're OpenAI credentials in the .env file in the api folder. Specify the service you which to use by providing AzureOpenAI or OpenAI to the GLOBAL__LLM_SERVICE variable.
GLOBAL__LLM_SERVICE="AzureOpenAI" # or "OpenAI"
AZURE_OPEN_AI__DEPLOYMENT_TYPE="chat-completion"
AZURE_OPEN_AI__CHAT_COMPLETION_DEPLOYMENT_NAME="xxx"
AZURE_OPEN_AI__TEXT_COMPLETION_DEPLOYMENT_NAME="xxx"
AZURE_OPEN_AI__ENDPOINT="https://xxx.openai.azure.com"
AZURE_OPEN_AI__API_KEY="xxx"
OPEN_AI__MODEL_TYPE="chat-completion"
OPEN_AI__CHAT_COMPLETION_MODEL_ID="gpt-4"
OPEN_AI__TEXT_COMPLETION_MODEL_ID="text-davinci-003"
OPEN_AI__API_KEY="xxx"
OPEN_AI__ORG_ID="xxx"