Tag-AI is a desktop application that uses artificial intelligence to automatically analyze and tag your photos, making them searchable by content without manual tagging. It can process standard image formats (JPG, PNG, etc.) as well as RAW formats from various camera manufacturers.
Tag-AI only requires internet connectivity during initial installation and setup. For ongoing use, you can choose between:
Tag-AI has been designed to handle large image libraries. The practical limits depend on your hardware, but libraries of 100,000+ images are supported with appropriate system resources.
No, Tag-AI never modifies your original images during scanning and tagging. The information is stored in Tag-AI's database. When using the "Push Tags to Image" feature, metadata is added to your image files but the actual image data remains unchanged.
No, Tag-AI is a one-time purchase. After buying a license key, you can use the software indefinitely without recurring payments.
Your Tag-AI license allows installation on up to two computers that you own (e.g., a desktop and a laptop).
If you're replacing a computer, you can deactivate your license on the old device using the License Manager (accessible via the API Test tool). Then you can activate it on the new device.
Tag-AI itself is proprietary software, but it relies on several open source components. You can view the licenses for these components in the application by going to Actions → Licenses.
Processing speed depends primarily on whether you have GPU acceleration properly configured. Without GPU acceleration, image processing can be 5-10x slower. Check that:
Processing speed varies widely depending on your hardware:
Currently, Tag-AI utilizes a single GPU. In systems with multiple GPUs, it will typically use the primary GPU.
Yes, Tag-AI works on Apple Silicon Macs (M1/M2/M3) and automatically uses the integrated GPU through the Metal API. No additional setup is required for GPU acceleration on these systems.
Tag-AI calculates a unique hash for each image file based on its contents. If the same image exists in multiple locations, it will only be processed once. If the file is moved, Tag-AI will update the path reference in its database.
Yes, you can stop a scan at any time by clicking the "Stop" button in the scan window. When you start a new scan later, Tag-AI will automatically skip files that have already been processed.
Tag-AI commits data to the database in batches, so most processed images will still be tagged. When you restart and scan again, it will automatically skip already-processed images and continue from where it left off.
To reprocess images with a different model:
The number of tags varies depending on the complexity of the image and the model used:
You can search for virtually anything visible in your images:
The AI models have certain limitations:
You can always add your own custom tags through the metadata editor.
Tag-AI supports several search patterns:
beach sunset finds images with both tags"red car" finds images with that specific tag"mountain landscape" snow finds images with both termsIn the default local processing mode, your images never leave your computer. If you choose to use the Gemini API option, resized versions of your images are sent to Google's servers for processing. You can always see which mode is active in your configuration.
Tag-AI itself does not collect or transmit any usage data or analytics. When using the local processing mode, all operations happen entirely on your device. The only external communication is for license validation, which does not involve your images or tags.
The Tag-AI database is not encrypted by default. If you're concerned about privacy, you can place the database on an encrypted drive or partition for added security.
Tag-AI can scan images on network drives and NAS systems, but performance will depend on your network speed. For optimal performance, we recommend:
Advanced users can configure Tag-AI to use alternative Ollama models:
ollama pull bakllavaTo reset Tag-AI to a fresh installation state: