Simple Agent Designer Documentation
Index

Provider Setup

This page provides step-by-step instructions for configuring each supported AI provider in Simple Agent Designer.

Accessing AI Provider Configuration

  1. Search for Agent Setup in the Tell Me search bar
  2. Choose AI Providers from the action bar
  3. The AI Providers list opens, showing all configured providers

The list displays three columns:

Column Description
Provider The AI provider type (read-only after creation)
Endpoint The API endpoint URL for this provider
Access Key The authentication key (displayed as masked text for security)

Providers Are Pre-Seeded on First Setup

The first time you open Agent Setup, Simple Agent Designer automatically creates a row for every supported provider except Azure OpenAI, each pre-filled with the correct default endpoint. A notification warns you that the API keys are still empty. In most cases you only need to:

  1. Open the provider you want to use
  2. Click the Access Key field — this opens that provider's API-key portal in your browser
  3. Create or copy a key and paste it into the field

Azure OpenAI is the exception: it is not seeded because its endpoint is unique to your Azure resource, so you must add it manually (see below).

When you click the masked Access Key field for a provider, the extension opens that provider's API-key portal automatically, so you rarely need to navigate there by hand.

Setting Up ChatGPT OpenAI

Prerequisites

Configuration

  1. In the AI Providers list, find or create the ChatGPT OpenAI entry
  2. Set the Endpoint to:
   https://api.openai.com/v1/chat/completions
  1. Click on the Access Key field
  2. A browser window opens to OpenAI API Keys
  3. Create a new API key in OpenAI (or copy an existing one)
  4. Paste the API key into the Access Key field

How It Connects

Simple Agent Designer sends requests to the ChatGPT OpenAI API using:

  • Authorization header: Bearer <your-api-key>
  • Content-Type: application/json
  • Standard OpenAI Chat Completions API format
  • Default model: gpt-5-mini (used when the agent's LLM Model field is left blank)

Setting Up Anthropic Claude

Prerequisites

Configuration

  1. In the AI Providers list, find or create the Anthropic entry
  2. Set the Endpoint to:
   https://api.anthropic.com/v1/messages
  1. Click on the Access Key field
  2. A browser window opens to Anthropic Settings
  3. Create a new API key in Anthropic (or copy an existing one)
  4. Paste the API key into the Access Key field

How It Connects

Simple Agent Designer sends requests to the Anthropic API using:

  • x-api-key header: <your-api-key>
  • anthropic-version header: 2023-06-01
  • Content-Type: application/json
  • Anthropic Messages API format (distinct from the OpenAI format)
  • Default model: claude-haiku-4-5 (used when the agent's LLM Model field is left blank)

Setting Up Azure OpenAI

Prerequisites

  • An Azure subscription
  • Azure OpenAI Service deployed in your Azure tenant
  • A model deployed in Azure AI Foundry (e.g., GPT-4)

Configuration

  1. In the AI Providers list, find or create the Azure OpenAI entry
  2. Set the Endpoint to your Azure OpenAI deployment URL, for example:
   https://your-resource.openai.azure.com/openai/deployments/your-deployment/chat/completions?api-version=2024-02-01
  1. Click on the Access Key field
  2. A browser window opens to Azure AI Foundry
  3. Copy your deployment's API key from the Azure portal
  4. Paste the API key into the Access Key field

How It Connects

Simple Agent Designer sends requests to Azure OpenAI using:

  • api-key header: <your-api-key>
  • Content-Type: application/json
  • Azure OpenAI Chat Completions API format

Finding Your Azure Endpoint

  1. Go to the Azure Portal
  2. Navigate to your Azure OpenAI resource
  3. Go to Keys and Endpoint to find your base URL
  4. Go to Model Deployments to find your deployment name
  5. Construct the full URL using the pattern above

Setting Up LM Studio

Prerequisites

  • LM Studio installed on an accessible machine
  • A model downloaded and loaded in LM Studio
  • The LM Studio server started and listening

Configuration

  1. In the AI Providers list, find or create the LM Studio entry
  2. Set the Endpoint to your LM Studio server URL, for example:
   http://localhost:1234/v1/chat/completions
  1. Set the Access Key (leave blank if LM Studio does not require authentication, or enter the configured key)

How It Connects

LM Studio provides an OpenAI-compatible API, so Simple Agent Designer communicates with it using the same format as ChatGPT OpenAI.

Network Considerations

  • If LM Studio runs on the same machine as Business Central, use localhost
  • If LM Studio runs on a different machine, use that machine's IP address or hostname
  • Ensure the port is accessible from the Business Central server
  • For on-premises BC, ensure firewall rules allow the connection

Setting Up the Additional OpenAI-Compatible Providers

The following providers all use the OpenAI Chat Completions format and authenticate with an Authorization: Bearer <your-api-key> header. For each one, the default endpoint is pre-seeded on first setup — you normally just need to add the API key. Click the Access Key field to open the provider's key portal, create or copy a key, and paste it in.

xAI Grok

  • Endpoint: https://api.x.ai/v1/chat/completions
  • API key portal: console.x.ai (opens when you click Access Key)
  • Default model: grok-4

Mistral

DeepSeek

Moonshot Kimi

Alibaba Qwen

  • Endpoint: https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions (international endpoint)
  • API key portal: bailian.console.alibabacloud.com
  • Default model: qwen-plus

Zhipu GLM

MiniMax

  • Endpoint: https://api.minimax.io/v1/chat/completions
  • API key portal: platform.minimax.io
  • Default model: MiniMax-M2

Setting Up Google Gemini

Prerequisites

  • A Google AI Studio account
  • A valid API key

Configuration

  1. In the AI Providers list, find or create the Google Gemini entry
  2. Confirm the Endpoint is:
   https://generativelanguage.googleapis.com/v1beta/models/
  1. Click on the Access Key field
  2. A browser window opens to Google AI Studio API Keys
  3. Create a new API key (or copy an existing one)
  4. Paste the API key into the Access Key field

How It Connects

Google Gemini uses its own JSON request/response format (not the OpenAI shape). Simple Agent Designer sends:

  • x-goog-api-key header: <your-api-key>
  • Content-Type: application/json
  • Default model: gemini-2.5-flash

Testing Your Configuration

After configuring a provider, test it by:

  1. Creating or selecting an agent that uses this provider
  2. Going to the Role Center and using the interactive chat
  3. Sending a simple message like "Hello, what model are you?"
  4. If you receive a response, the provider is configured correctly

Troubleshooting Provider Configuration

Issue Possible Cause Solution
No response from AI Invalid endpoint URL Verify the endpoint URL is correct and accessible
Authentication error Invalid or expired API key Regenerate the API key in the provider's console
Timeout errors Network connectivity issue Ensure your BC server can reach the provider's endpoint
Rate limit errors Too many requests The system retries automatically; consider upgrading your provider tier
Empty responses Model not available Verify the model name in your agent configuration matches an available model

Tips and Best Practices

  • Keep your API keys secure — they are stored in the Business Central database with masking on the UI
  • Set up at least two providers for redundancy
  • Monitor your provider dashboards for usage and billing
  • Test new provider configurations with a simple chat before running production workloads
  • For Azure OpenAI, ensure your deployment's content filtering settings are appropriate for your use case