Skip to main content

Robi AI Prerequisites

Before setting up Robi AI, ensure that you have met the following prerequisites.

AI License​

You must have an AI License to access Robi AI features.

To verify that your license includes AI, navigate to the System Details widget in the Universal Controller. If your license includes AI, you will see AI: True listed in the license information.

Azure OpenAI​

Robi AI uses Azure OpenAI for its chat and embedding models. You must have:

  • An active Azure subscription with the Azure OpenAI service provisioned
  • Two OpenAI Deployments: one for the chat and one for doc embeddings. You must note the Deployment Name of each.
  • An API key for your Azure OpenAI resource
  • The Endpoint URL for your Azure OpenAI resource

The following sections will guide you through setting up Azure OpenAI and getting the API Key and Endpoint URL needed to configure the AI Service.

info

Robi AI officially supports Azure OpenAI only. While other model providers may be technically configurable, they are not supported.

Models​

Robi AI requires two Azure OpenAI models deployed in your Azure environment:

Required ModelModel IDPurpose
GPT-4.1gpt-4.1 (2025-04-14)Chat / Completions
Text Embedding 3 Smalltext-embedding-3-smallEmbeddings

Both models are Foundry Models sold directly by Azure.

For full model specifications and region availability, see the Microsoft documentation:

Azure OpenAI Resource vs. Azure AI Foundry​

Microsoft offers two paths to deploy OpenAI models on Azure. Either path works for the models we require. The right choice depends on your existing environment and broader AI needs.

An Azure OpenAI resource gives you direct, focused access to OpenAI models via REST API and SDK.

Choose this if:

  • You only need the OpenAI models listed above (gpt-4.1 and text-embedding-3-small).
  • You want a straightforward setup with minimal overhead.
  • You are already using an Azure OpenAI resource for other workloads.

How to set up:

  1. In the Azure portal, create an Azure OpenAI resource (or reuse an existing one) in a supported region.
  2. Navigate to the resource > Model deployments > Deploy model.
  3. Deploy gpt-4.1 - select the Standard or Global Standard deployment type.
  4. Deploy text-embedding-3-small - same process.
  5. Note your Endpoint URL and API Key, and the Deployment Name of each model.

Option B: Azure AI Foundry (Project-based)​

Azure AI Foundry (formerly Azure AI Studio) is a broader platform that embeds Azure OpenAI as one of its components. It adds multi-model access, agent orchestration, evaluation tooling, and lifecycle management.

Choose this if:

  • You already have an Azure AI Foundry project set up.
  • You plan to use additional models beyond OpenAI (e.g., Meta Llama, Mistral, DeepSeek, Cohere).
  • You want a unified platform for experimentation, fine-tuning, monitoring, and agent workflows.

How to set up:

  1. Sign in to Microsoft Foundry and open (or create) a Foundry project.
  2. Go to Discover > Models, search for gpt-4.1.
  3. Select Deploy > Default settings (or Custom settings to pick a specific region/deployment type).
  4. Repeat for text-embedding-3-small.
  5. From the deployment detail page, note your Endpoint URL and API Key, and the Deployment Name of each model.
info

Because both models are "sold directly by Azure," the deployment and API experience is identical regardless of whether you use a standalone Azure OpenAI resource or a Foundry project. The API endpoint format and authentication work the same way.

Locating the API Key, Endpoint URL, and Deployment Name​

If you are having trouble locating your API Key, Endpoint URL, or Deployment Name, there are two simple ways to find them in Azure.

Foundry​

In Azure Foundry, go to Deployments, then view the details of your chat (gpt-4.1) model.

  • The Endpoint URL is the base URL of the Target URI.
  • Your API Key is in the Key field
  • Your Deployment Name is the Name in the "Deployment info" section

You should also check the Deployment Name of the embedding (text-embedding-3-small) model.

Azure Portal​

You can find the same information in the Azure Portal. For each deployment, go to Resource Management > Keys and Endpoint.

  • As shown above, there are 2 keys provided (for key rotation). You can use KEY 1 as your API Key, however, either would work.
  • Note the Endpoint