Skip to main content

​
Supported Providers

Sorty supports multiple AI providers, from cloud-based services to on-device models:

OpenAI

GPT-4, GPT-4 Turbo, GPT-3.5Supports vision mode for image analysis.

Anthropic

Claude 3.5 Sonnet, Claude 3 Opus/HaikuStrong reasoning, detailed explanations.

Apple Intelligence

On-device foundation modelsPrivate, no data leaves your Mac. Requires macOS 15.1+, M-series chip.

Ollama

Local LLMs (Llama, Mistral, etc.)Run models locally for complete privacy.

GitHub Copilot

GitHub Copilot LLM accessUse your existing Copilot subscription.

OpenRouter

Multi-provider gatewayAccess to 100+ models through one API.
All providers implement the AIClientProtocol, making them interchangeable.

​
Provider Factory

Sorty uses a factory pattern to instantiate the correct client: 
public struct AIClientFactory {
    public static func createClient(config: AIConfig) throws -> AIClientProtocol {
        switch config.provider {
        case .openAI, .groq, .openAICompatible, .openRouter, .ollama, .gemini:
            return OpenAIClient(config: config)
            
        case .githubCopilot:
            return GitHubCopilotClient(config: config)
            
        case .anthropic:
            return AnthropicClient(config: config)
            
        case .appleFoundationModel:
            if AppleFoundationModelClient.isAvailable() {
                return AppleFoundationModelClient(config: config)
            }
            throw AIClientError.apiError(
                statusCode: 501,
                message: "Apple Intelligence is not supported on this version of macOS."
            )
        }
    }
}

​
Configuration

​
OpenAI

1

Get API Key

Sign up at platform.openai.com and create an API key.
2

Configure in Sorty

Navigate to Settings → AI Provider and select OpenAI.
  • API URL: https://api.openai.com/v1
  • API Key: Your OpenAI API key
  • Model: gpt-4-turbo-preview (recommended)
3

Enable Vision (Optional)

Toggle Enable Vision to use GPT-4 Vision for image analysis.

​
Anthropic

1

Get API Key

Sign up at console.anthropic.com and create an API key.
2

Configure in Sorty

Navigate to Settings → AI Provider and select Anthropic.
  • API URL: https://api.anthropic.com/v1
  • API Key: Your Anthropic API key (starts with ant-api-)
  • Model: claude-3-5-sonnet-20241022 (recommended)

​
Apple Intelligence

Requires macOS 15.1+ (macOS Sequoia) and an M-series chip (M1/M2/M3/M4).
1

Enable Apple Intelligence

Go to System Settings → Apple Intelligence & Siri and enable Apple Intelligence.
2

Configure in Sorty

Navigate to Settings → AI Provider and select Apple Foundation Models.No API key required - models run on-device.
  • 100% On-Device: Nothing leaves your Mac
  • No API Costs: Free to use
  • No Network Required: Works offline
  • No Rate Limits: Process unlimited files
Limitations:
  • Smaller context window than cloud models
  • May be slower than server-side GPUs
  • No vision mode support (yet)

​
Ollama (Local Models)

1

Install Ollama

Download from ollama.ai and install.
# macOS/Linux
curl -fsSL https://ollama.ai/install.sh | sh
2

Pull a Model

ollama pull llama3.2
# or
ollama pull mistral
# or
ollama pull qwen2.5
3

Start Ollama Server

ollama serve
The server runs on http://localhost:11434 by default.
4

Configure in Sorty

Navigate to Settings → AI Provider and select Ollama.
  • API URL: http://localhost:11434/v1
  • Model: llama3.2 (or your pulled model)
  • Requires API Key: OFF

​
GitHub Copilot

1

Authenticate

Navigate to Settings → AI Provider and select GitHub Copilot.Click Authenticate to sign in with GitHub.
2

Verify

Sorty will use your existing Copilot subscription.
Requires an active GitHub Copilot subscription ($10/month for individuals).

​
OpenRouter

1

Get API Key

Sign up at openrouter.ai and create an API key.
2

Configure in Sorty

Navigate to Settings → AI Provider and select OpenRouter.
  • API URL: https://openrouter.ai/api/v1
  • API Key: Your OpenRouter key
  • Model: See available models

​
Streaming Support

All providers support streaming responses for live progress updates: 
public protocol AIClientProtocol {
    var streamingDelegate: StreamingDelegate? { get set }
    
    func analyze(
        files: [FileItem],
        customInstructions: String?,
        personaPrompt: String?,
        temperature: Double?
    ) async throws -> OrganizationPlan
}

public protocol StreamingDelegate {
    func didReceiveChunk(_ chunk: String)
    func didComplete(content: String)
    func didFail(error: Error)
}
  1. Client sends request with streaming enabled
  2. Server responds in chunks via Server-Sent Events (SSE)
  3. Delegate receives chunks and updates UI in real-time
  4. Progress insights are extracted from >> prefixed lines
  5. JSON output is parsed when complete

​
Error Handling

Sorty provides detailed, user-friendly error messages:
Authentication failedYour API key may be invalid or expired. Check your credentials in Settings.
API keys are automatically redacted from error messages for security.

​
Vision Support

​
Compatible Providers

ProviderVision SupportModels
OpenAI✅ Yesgpt-4-vision-preview, gpt-4-turbo, gpt-4o
Anthropic✅ Yesclaude-3-opus, claude-3-sonnet, claude-3.5-sonnet
Ollama⚠️ Limitedllava, bakllava (experimental)
Apple Intelligence❌ Not yet-
GitHub Copilot❌ No-

​
How Vision Analysis Works

1

Image Detection

Sorty identifies image files by extension:
private let visionImageExtensions: Set<String> = 
    ["jpg", "jpeg", "png", "heic", "webp"]
2

Image Preprocessing

Images are resized and encoded as base64 (max 1024x1024 for efficiency).
3

Multimodal Request

Images are sent alongside filenames to the AI:
func analyzeWithImages(
    files: [FileItem],
    imageData: [String: Data],
    customInstructions: String?,
    personaPrompt: String?,
    temperature: Double?
) async throws -> OrganizationPlan
4

Vision Insights

AI analyzes image content and suggests organization based on:
  • Objects and scenes
  • Text within images
  • Visual similarity
  • Context clues
Vision analysis significantly increases API costs and processing time. Use selectively.

​
Deep Scan Support

​
Provider Compatibility

extension AIProvider {
    public var supportsDeepScan: Bool {
        switch self {
        case .openAI, .anthropic, .openAICompatible, .openRouter:
            return true
        case .ollama, .githubCopilot, .appleFoundationModel:
            return false
        }
    }
}
  • Ollama: Local models have smaller context windows
  • Apple Intelligence: API limitations
  • GitHub Copilot: Designed for code completion, not document analysis

​
Timeout & Retry

Configure request timeouts in Settings → Advanced: 
// Default timeout
let timeout: TimeInterval = 120.0 // 2 minutes

// For large batches
let extendedTimeout: TimeInterval = 300.0 // 5 minutes
Increase timeout for slow providers or large file batches.

​
API Key Security

API keys are stored securely in macOS Keychain: 
// Keys are never stored in UserDefaults or plain text
Keychain.save(apiKey, for: "sorty.api.key")

// Privacy Mode blurs keys in UI
if privacyModeEnabled {
    displayedKey = "sk-••••••••••••••••"
}

File Organization

Learn how the AI organizes your files

Personas

Customize AI behavior for different workflows

The Learnings

Train the AI on your preferences