Train, Deploy & Run
AI Models On-Device

Dataset Management

Import datasets from multiple sources—local folders, videos, Parquet files, DICOM files, or directly from HuggingFace. Organize by class and manage your training data with powerful statistics.

• Local folder import
• Video/movie file import
• Parquet file support
• DICOM file import
• HuggingFace dataset download
• Automatic class detection
• Dataset statistics & visualization
• Train/validation split management

Model Training

Train CoreML models using CreateML with full control over training parameters and iterations. Support for classification tasks.

• CreateML integration
• Classification training
• Custom preprocessing configurations
• Custom training iterations

Model Import

Import models in multiple formats. Support for CoreML (.mlmodel, .mlpackage) and ONNX (.onnx) models with automatic compilation and optimization. Download models directly from HuggingFace with automatic conversion.

• CoreML model import
• ONNX model import
• HuggingFace model download
• Automatic server-side conversion
• Automatic model compilation
• Model validation & optimization

HuggingFace Integration

Seamlessly download datasets and models from HuggingFace. Models are automatically converted to ONNX or CoreML format on our server, ready to use in your projects. Browse thousands of pre-trained models and datasets.

• Download datasets from HuggingFace
• Download models from HuggingFace
• Automatic model conversion (ONNX/CoreML)
• Server-side conversion service
• Browse popular models & datasets
• Search by task type & license
• Direct import to your projects

Note: Not all models from HuggingFace will work. Model compatibility depends on architecture and conversion support.

Image Classification

Run inference on images with your trained models. Get instant predictions with confidence scores and detailed analytics.

• Batch image processing
• Multi-model inference
• Confidence scores
• Performance metrics

Object Detection

Detect and locate objects in images with bounding boxes. Perfect for YOLO models and custom detection tasks.

• YOLO model support
• Bounding box visualization
• Multi-object detection
• Export annotations

Segmentation

Pixel-level segmentation for advanced ML tasks. Run segmentation models with visual overlays.

• Semantic segmentation
• Mask visualization
• Export capabilities

Advanced Analytics

Comprehensive metrics and visualizations. Confusion matrices, training curves, and detailed performance reports.

• Confusion matrices
• Training metrics
• Validation analytics
• Export to JSON/CSV/PDF

Video Processing

Process videos frame-by-frame with AI models. Export annotated videos with bounding boxes, labels, and segmentation overlays.

• Frame-by-frame analysis
• Custom frame intervals
• Video overlay export
• Bounding boxes & labels on video
• Segmentation mask overlays
• Real-time preview

Live Camera Inference

Real-time AI inference using your iPhone camera. Run classification, detection, and segmentation models live with instant results.

• Live classification
• Real-time object detection
• Live segmentation overlay
• On-device processing
• Frame-by-frame updates

DICOM Support

Import and process DICOM medical imaging files. Extract frames, view multi-frame sequences, and run AI inference on medical images.

• DICOM file import
• Multi-frame extraction
• DICOM viewer
• Frame-by-frame analysis
• DICOM header reading
• Medical image processing

Automation & Triggers

Automate actions when objects are detected or images are classified. Configure triggers to run shortcuts, send notifications, open webpages, or provide haptic feedback—all via App Intents.

• Configurable detection triggers
• Classification-based triggers
• Run Shortcuts app actions
• Send notifications with custom messages
• Open URLs automatically
• Haptic feedback on detection
• Confidence threshold configuration
• All configurable in Settings