AI AGENT DEVELOPMENT ON AZURE

In today's rapidly evolving business landscape, artificial intelligence is no longer a luxury but a necessity for organizations of all sizes. Failure to embrace AI risks being left behind or even becoming obsolete. A practical starting point could be implementing an AI agent capable of answering employee inquiries regarding expense claims. This can then evolve into more sophisticated solutions, such as automating the entire expense claim process or developing an intelligent agent that can connect to organizational databases and provide natural language answers to data-related questions. Such implementations promise to eliminate paperwork, streamline workflows, and significantly boost overall efficiency. Having recently completed the Azure AI Foundry skill fest challenge, I gained hands-on experience in this domain, including building a basic yet functional expense claim question-answering agent, which I'd be happy to share.

AZURE AI FOUNDRY

Azure AI Foundry is Microsoft's platform for building, deploying, and managing AI applications. It provides developers with access to AI models, tools, and infrastructure to create intelligent solutions. The platform supports a variety of AI models, including those from Azure OpenAI Service and other providers, allowing users to customize and fine-tune models for their specific needs.

Azure AI Foundry is designed to help organizations scale AI applications efficiently while ensuring responsible AI practices. It offers features like model evaluation, fine-tuning, and deployment options, making it a comprehensive environment for AI development.

•In this exercise, the Azure AI Agent service tools in the Azure AI Foundry portal was used to create a simple AI agent that answers questions about expense claims.

Step 1

Create an Azure AI Foundry Project

In a web browser open the azure AI foundry portal with https://ai.azure.com sign in using your Azure credentials and click on + create project

In the Create a project wizard, enter a valid name for your project and if an existing hub is suggested, choose the option to create a new one. Then review the Azure resources that will be automatically created to support your hub and project, click on customize to create a new hub for your project, fill the required field, click the help me choose section to automatically suggest a location for you and click on next.

Review the resources you have created and click on create

Review the project page in Azure AI Foundry portal

Step 2

Deploy a generative AI Model

In the pane on the left for your project, in the My assets section, select the Models + endpoints page. In the Models + endpoints page, in the Model deployments tab, in the + Deploy model menu, select Deploy base model

Search for the gpt-40 model and click on it then click on confirm

Customize the model with the required information, select the latest model version

Step 3

Create an AI Agent

build a simple agent that answers questions based on a corporate expenses policy, assume  you have the expense policy in your local PC, the agent will search the document and produce the required answer to the question. Return to the browser tab containing the Azure AI Foundry portal, in the navigation pane on the left, in the Build and customize section, select the Agents page, If prompted, select your Azure OpenAI service resource and click on let's go.

Select your new agent. Then, in the Setup pane for your new agent, set the Agent name to ExpenseAgent, ensure that the gpt-4o model deployment you created previously is selected, and set the Instructions to Answer questions related to expense claims. Further down in the Setup pane, next to the Knowledge header, select + Add. Then in the Add knowledge dialog box, select Files

Then in the Add knowledge dialog box, select Files

in the Adding files dialog box, create a new vector store named expenses_vector_store,in step 1, select the local file to be uploaded in step 2 then upload and save the file in step 3. in the Setup pane, in the Knowledge section, verify that Expenses_Vector_Store is listed and shown as containing 1 file.

Step 4

Test your Agent

At the top of the Setup pane for your agent, select Try in playground.

in the playground, enter the prompt ‘what is the maximum I can claim for hotel accommodation’ and review the agent’s response - which should be based on information in the expenses policy document you added as knowledge to the agent setup

Review the agent response and observe it is based on the information in the expense policy document uploaded as a knowledge file in the agent setup.

You can contact us for better understanding and for your AI DEPLOYMENT and also learn more on https://microsoftlearning.github.io/mslearn-ai-agents/Instructions/01-agent-fundamentals.html

Thanks