Advancing AI with Azure AI Foundry
Datagaps Suite 3.0.0:
In version 3.0.0, we introduced advanced AI features into our platform by using embedded AI models and deploying them through Azure OpenAI. For deployment, we relied on local virtual machines, which made the setup more complex. While this approach allowed us to successfully bring intelligence into the system, it also had limitations in terms of flexibility and scale.
As data volumes and business requirements grew, the setup struggled to keep up, leading to challenges in performance, scalability, and long-term efficiency.
Key Challenges in 3.0.0:
- Performance issues while handling large workloads.
- Limited storage space.
- Deployment was fully manual and required running on a virtual machine.
- Monitoring usage and performance was difficult, with limited visibility.
- Costs were higher due to the need for dedicated infrastructure.
- Expanding or adding new AI use cases required extra effort and setup.
Moving Forward: Datagaps Suite 4.0.0
With version 4.0.0, we are moving to Azure AI Foundry, Microsoft’s enterprise AI platform. This shift removes the earlier challenges and sets the stage for long-term growth.
What is Azure AI Foundry:
Azure AI Foundry is a unified Azure platform-as-a-service offering for enterprise AI operations, model builders, and application development. This foundation combines production-grade infrastructure with friendly interfaces, enabling developers to focus on building applications rather than managing infrastructure.
Why Azure AI Foundry?
- Dedicated and Secure: Each customer gets their own non-shared environment, ensuring complete data privacy.
- Simple Project Flow: We can create a project inside AI Foundry, and within that project, set up multiple agents for different needs.
- Easy Monitoring: AI Foundry provides built-in tools to monitor usage and performance.
- Cost Control: Operates on a pay-as-you-go model, so you only pay for what you use.
- Future Ready: It prepares us for upcoming AI innovations and ensures compliance with enterprise standards.
Steps to Create an Azure AI Foundry Resource and Project
1. Login to the Azure portal.
2. Search for “AI Foundry” in the search bar and open the service.
3. Under "Use with AI Foundry", select "AI Foundry".
4. Click on "Create".
5. Basics section:
-> Select the Subscription.
-> Select an existing Resource group from the dropdown, or create a new one.
-> Enter a Foundry Name. (no spaces, must be globally unique)
-> Select the Region.
-> Enter a Project Name.
Click ‘Next’.
6. Network section – Select preferred inbound access:
-> All networks – Any network, including the internet, can access this resource.
-> Selected networks – Configure network security for your Azure AI Foundry resource.
-> Disabled – No networks can access this resource, only private endpoint connections can be used.
Click ‘Next’.
7. Identity section – Choose the identity type:
-> System-assigned.
-> User-assigned.
Click ‘Next’.
8. Encryption section – If required, enable "Encrypt data using a customer-managed key".
Click ‘Next’.
9. Tags – (Optional) Add tags to organize and manage your resource.
Click ‘Next’.
10. Review + Create – Review all configuration details
Click "Create".
If the deployment is completed, then we can see that the creation got completed:
Deploy Model:
Once the AI Foundry got created then deploy the model.
- Open the Foundry that we have created.
- Click on the ‘Go to Azure AI Foundry Portal’
- In Left side menu, click on ‘Models + endpoints’ in My assets section:
- Click on ‘Deploy model’ -> ‘Deploy base model’
- Select the model to get deployed:
- Click on ‘Confirm’:
- Click on ‘Deploy’:
- Once it got deployed, we can see the Deployment info like below:
Prerequisites
- Active Azure subscription.
- Choose region.
- Existing resource group or create new one.
- Networking and encryption.
Required Roles
Azure-level
- Contributor (Recommended) — at the resource group level where AI Foundry and related services will be created.
- Owner (Optional) — only if the user must assign permissions to others.
AI Foundry-level
- AI Foundry Contributor — allows creating projects, agents, models, deployments inside the workspace.
- AI Foundry Owner (Optional) — if the user also needs to manage workspace access.