AI Settings - Azyrom Help

Azyrom Designer supports multiple AI providers for code generation in Make Mode. Configure your preferred provider and API keys to enable AI-powered features.

Supported AI Providers

Choose from four AI providers:

Anthropic (Claude)

Model: claude-sonnet-4-20250514
Best for: High-quality code generation with excellent reasoning
Requires: Anthropic API key
Get key: console.anthropic.com

OpenAI (GPT)

Model: gpt-4o
Best for: Fast responses with broad knowledge
Requires: OpenAI API key
Get key: platform.openai.com

Ollama (Local)

Model: llama3.2 (customizable)
Best for: Privacy-focused users, no API costs
Requires: Ollama installed locally
Install: ollama.ai (free)

Apple Intelligence (macOS Only)

Model: apple-on-device
Best for: Privacy, no API costs, works offline
Requires: macOS 26+ on Apple Silicon
Cost: Free (100% on-device inference)
Limitations: Text-only (no image input), smaller context window than cloud models
Availability: Automatically detected and enabled if your system supports it

Provider Comparison

Feature	Anthropic	OpenAI	Ollama	Apple
Cost	Pay per token	Pay per token	Free	Free
Privacy	Cloud-based	Cloud-based	100% local	100% on-device
Internet Required	Yes	Yes	No	No
Image Input	✅ Yes	✅ Yes	❌ No	❌ No
Code Quality	Excellent	Very Good	Good	Good
Setup	API key	API key	Install app	Auto (if available)

Opening AI Settings

Access the AI Settings dialog:

Open Make Mode
Click the Settings icon in the Make Mode toolbar
Or use the menu: File → AI Settings

Configuring Your API Key

Step 1: Select Provider

Use the segmented button at the top to choose your AI provider:

Anthropic - Claude models (cloud-based, excellent quality)
OpenAI - GPT models (cloud-based, very good quality)
Ollama - Local models (privacy-focused, free)
Apple - Apple Intelligence (on-device, macOS 26+ only)

Step 2: Enter API Key

For Anthropic or OpenAI:

Paste your API key in the appropriate field
Click the Save icon (💾) or press Enter
Keys are stored securely on your local device
Use the eye icon (👁️) to toggle key visibility

Step 3: Verify Configuration

The status indicator shows:

Green checkmark - "{Provider} Ready" - Configured correctly
Orange warning - "{Provider} Not Configured" - API key required

Ollama Configuration

To use Ollama (local AI):

Download and install Ollama from ollama.ai
Start Ollama (runs on http://localhost:11434 by default)
Pull a model: ollama pull llama3.2
In Azyrom AI Settings, select Ollama provider
Verify the URL is correct (default: http://localhost:11434)
Optionally, change the model name to use a different installed model

Benefits: Free, runs locally, no API costs, privacy-focused

Apple Intelligence Configuration

Apple Intelligence is Apple's on-device AI available on macOS 26 and later:

System Requirements

Operating System: macOS 26.0 or later
Hardware: Apple Silicon (M1, M2, M3, M4, or later)
Xcode: Built with Xcode 26+ and FoundationModels framework

Setup

Apple Intelligence requires no configuration if your system meets the requirements:

Ensure you're running macOS 26+ on Apple Silicon
In Azyrom AI Settings, select Apple provider
If available, you'll see "Apple Intelligence Ready" with a green checkmark
If not available, you'll see "Requires macOS 26+ on Apple Silicon"

Benefits

Privacy: 100% on-device processing, no data sent to servers
Cost: Completely free, no API key required
Offline: Works without internet connection
Fast: Low latency on Apple Silicon

Limitations

Text Only: Does not support image input (unlike Anthropic/OpenAI)
Context Size: Smaller context window than cloud models
Capability: Good quality but not as advanced as Claude Sonnet 4 or GPT-4o
Platform: macOS 26+ only (not available on older systems)

When to Use Apple Intelligence

When you want 100% privacy (no cloud processing)
When working offline or with slow internet
For quick iterations where cost is a concern
When API rate limits are reached on other providers

When to Use Cloud Providers Instead

When you need image input support (design analysis)
When you need the highest code quality
When you need larger context windows for complex projects
When you're on an older macOS version or Intel Mac

Security & Storage

Local Storage: API keys are stored in your device's SharedPreferences
Not Synced: Keys stay on your device, not synced to cloud
Not Shared: Keys are never sent anywhere except the chosen AI provider
Masked Display: Keys are masked when shown (e.g., "sk-ant-••••1234")

Switching Providers

You can switch between providers at any time:

Open AI Settings
Click the provider you want to use
Ensure that provider's API key is configured
Close the dialog - your selection is saved automatically

All three providers can be configured simultaneously. Switch between them based on your needs.

Troubleshooting

API Key Not Working

Verify key is correct (copy-paste to avoid typos)
Check key has sufficient credits/quota
Ensure key has proper permissions
Try regenerating a new key from the provider's console

Ollama Connection Failed

Verify Ollama is running: ollama list
Check URL is correct (default: http://localhost:11434)
Ensure firewall allows local connections
Try: ollama serve to start the service

Apple Intelligence Not Available

Check macOS version: System Settings → General → About (must be 26.0+)
Verify you have Apple Silicon: About This Mac → Chip (must show M1/M2/M3/M4)
Update macOS if you're on an older version
If on Intel Mac, Apple Intelligence is not supported (use Anthropic/OpenAI/Ollama instead)

Request Failed / Rate Limit

Check your API usage quota on provider's dashboard
Wait a few minutes if you hit rate limits
Consider switching to another configured provider temporarily

Best Practices

Configure Multiple Providers: Set up 2-3 providers as backups
Privacy-First Development: Use Apple Intelligence or Ollama for sensitive projects
Cost Management: Use Apple/Ollama for testing, cloud providers for production
Monitor Usage: Check API usage regularly to avoid surprise costs
Image Analysis: Use Anthropic or OpenAI when you need to analyze design screenshots
Secure Your Keys: Never share API keys publicly
Rotate Keys: Regenerate keys periodically for security