How to Configure AI Provider

Contents:
Introduction
Configuring Ollama (Local)
Configuring Azure OpenAI
Configuring Claude (Anthropic)
Configuring ChatGPT (OpenAI API)
Troubleshooting and Best Practices

Introduction

This guide explains how to configure an AI Provider for your Dovestones application.
The system supports the following providers:

• Ollama (Local AI)
• Azure OpenAI
• Claude (Anthropic)
• ChatGPT (OpenAI API)

Note: You can configure only one AI provider at a time.

Before You Start

Make sure:

• You are running the latest supported version of the product.
• You have access to the AI Configuration settings in the application.
• You have an active account for the selected provider (if required).

Configuring Ollama (Local AI Provider)

Ollama allows you to run AI models locally on your machine (no external API required).
Recommended for:

• Offline environments
• Maximum privacy
• No API costs

Step 1 — Install Ollama

Go to the official website: https://ollama.com
Download and install Ollama for your OS.
After installation, verify it is running: ollama --version
If installed correctly, it will return the installed version.

Step 2 — Pull an AI Model

You must download at least one model.
Example (recommended lightweight model): ollama pull llama3.1:8b
To verify installed models: ollama list

Step 3 — Start Ollama Server

Ollama runs automatically as a background service.
By default it listens on: http://localhost:11434
You can test it: curl http://localhost:11434/api/tags
If working, it will return available models in JSON format.

Step 4 — Configure in Dovestones Application

Open AI Settings and set:
Provider Type: Ollama (Local)

⚠ Important Notes

• Ensure port 11434 is not blocked by firewall.
• Large models require significant RAM.
• If you receive “unable to allocate CPU buffer”, use a smaller model.

Configuring Azure OpenAI

Use this option if you have an Azure subscription.

Step 1 — Create Azure OpenAI Resource

1. Login to Azure Portal.
2. Create new resource → Azure OpenAI.
3. Deploy model (example: gpt-4o, gpt-4, gpt-35-turbo).

After deployment, note:

• Endpoint URL
• API Key
• Deployment Name

Step 2 — Get Required Information

From Azure Portal → Keys and Endpoint
Example:
Endpoint: https://your-resource-name.openai.azure.com
Key: xxxxxxxxxxxxxxxxxxxxxxxx
Deployment Name: gpt-4o

Step 3 — Configure in Application

Provider Type: AzureOpenAI

Save and test.

⚠ Common Issues

• 401 Unauthorized → wrong key
• 404 Deployment not found → incorrect deployment name
• Timeout → firewall/VNET restrictions

Configuring Claude (Anthropic)

Claude requires an Anthropic API key.

Step 1 — Create Anthropic Account

Go to: https://console.anthropic.com
Create account and generate API key.

Step 2 — Select Model

Example models:

• claude-3-opus
• claude-3-sonnet
• claude-3-haiku

Step 3 — Configure in Application

Provider Type: ClaudeAI

Save and test.

⚠ Billing Warning

If you receive: Your credit balance is too low to access the Anthropic API
You must upgrade your plan or purchase credits.

Configuring ChatGPT (OpenAI API)

This uses the official OpenAI API.

Step 1 — Create OpenAI API Key

Go to: https://platform.openai.com
Navigate to:

API Keys → Create new key
Copy and store securely.

Step 2 — Choose Model

Examples:

• gpt-4o
• gpt-4
• gpt-3.5-turbo

Step 3 — Configure in Application

Provider Type: ChatGpt

Save and test.

⚠ Common Errors

• 401 Unauthorized → Invalid API key
• 429 Rate limit exceeded → Upgrade plan or reduce request frequency
• Invalid request error → Check model name

Troubleshooting

If AI does not respond:

• Check logs
• Validate API key
• Confirm model name
• Check firewall
• Verify endpoint URL

Best Practices

• Use Ollama for secure/offline environments.
• Use Azure OpenAI for enterprise production systems.
• Use ChatGPT for simplicity and quick setup.
• Use Claude for advanced reasoning tasks.