Documentation

Welcome to the official AutoStock documentation! AutoStock is an enterprise-grade desktop automation application designed specifically for high-volume stock contributors, photographers, and illustration agencies. It completely takes the manual work out of managing, qualifying, naming, and generating stock assets.

To get started, simply launch the desktop client, configure your connection settings in the settings panel, and drop your local folder of files into the workspace. Let's look at all of the features below!

The Generate Metadata feature is the core heart of AutoStock. Whenever you drag and drop any visual asset—whether a flat photo, a video sequence, or a vector graphic—our multi-modal AI immediately reads and analyzes the visual subjects, mood, context, and elements.

Instead of you wasting time thinking of search tags, the AI instantly drafts highly optimized, descriptive titles, detailed categories, and up to 50 keywords that precisely align with standard stock agency metadata requirements.

AutoStock provides broad support for multiple asset file types:

• Raster Images: High-speed subject and color depth analysis for JPEGs and PNGs.
• Vector Graphics: Analyzing boundaries, subjects, and compatibility for EPS and SVG illustrations.
• Video Files: Multi-frame scanning to extract context and high-action metadata from MP4s.

The AI Asset Renamer handles your file naming by translating raw visual context into search-optimized filenames. Instead of saving assets as generic camera numbers, the AI automatically analyzes the subject matter and builds SEO-friendly filenames using custom nomenclatures.

You can configure naming structures like {keyword}-{seq} or {title} in your settings screen to transform your folder structures instantaneously.

For example, setting a simple template like {keyword}-{seq} will rename a folder of raw images into travel-scenery-001.jpg and travel-scenery-002.jpg in seconds.

The Content Qualifier acts as your first-line QA system. It leverages deep filtering algorithms to scan files for commercial viability and pixel-perfect technical guidelines.

It runs automatic checks to identify image blur, layout issues, bad boundaries, or any intellectual property logos that could trigger platform rejection, sorting commercial vs editorial assets automatically.

AutoStock features a powerful, state-of-the-art Local AI Mode that enables you to execute compute-heavy neural pipelines directly on your device. By leveraging models running locally, you can process thousands of visual assets with zero cloud costs, complete offline privacy, and no dependency on external network speeds.

Local AI utilizes high-performance inference servers running on your local machine, such as Ollama or LM Studio. It operates seamlessly across three primary tools in the AutoStock suite:

System Prerequisites & Hardware Scaling

Because local AI models execute neural parameters directly on your system, performance is closely tied to your device's hardware specifications:

• Hardware Threshold: A minimum of 8GB unified system RAM and 4GB+ dedicated VRAM (NVIDIA RTX or Apple Silicon M-series) is highly recommended. Low-end architectures are automatically flagged in the UI controls panel.
• Model Recommendations: For the ultimate balance between accuracy, system resources, and inference speed, we recommend using gemma3:4b or gemma3:12b.

Currently Supported Models

AutoStock has official, first-class native support and optimization profiles for the following two model architectures under Local AI Mode:

AutoStock supports importing prompt lists directly using standard CSV files. This allows high-volume generative artists and design studios to orchestrate bulk, hands-free automation runs on Whisk or Google Labs Flow without manually pasting Sheets URLs.

However, to ensure our multi-threaded automation engine parses your queues flawlessly, your CSV files must adhere to strict structural constraints.

Strict Structural Rules

Your CSV must follow the single-column format. The critical rule: all prompts and negative prompts must be combined into single cell values. No spreadsheet cell can contain multiple separate prompts.

Negative Prompt Prefixes

Negative prompts must be written in the same cell as the main prompt, combined together using one of our officially supported syntax tags:

Download Example Template

To ensure zero parsing errors and get started instantly, click below to download our pre-validated CSV template. You can open it in Excel, Notepad, or Google Sheets:

The Whisk Automation module allows you to bridge external prompts and layouts directly with image generation pipelines.

It automates the copy-paste prompt formatting process from Whisk, letting you orchestrate consistent, stylized visual layouts without repetitive manual edits.

The Flow Automation workspace is designed for maximum speed and generative control. It runs concurrent image generation threads directly mapped to high-speed batch lists.

By grouping your text prompt lists, aspect ratios, and seed controls, Flow Automation generates thousands of visual assets concurrently with minimal user interaction.

The Recraft Automation pipeline integrates natively with the Recraft illustration engine.

It is optimized for high-volume vector illustrations, modern graphic icons, and stylized flat designs. It automatically manages custom style seeds and aspect scales to output consistent, premium vector-ready assets.

The Ideogram Automation feature runs batches through Ideogram's state-of-the-art models, which are highly celebrated for their clean typographic rendering.

If you are generating stock assets with typography, clean lettering, or text layout elements, Ideogram Automation renders flawless, sharp text layouts without letter distortions.

The Midjourney Automation engine manages Midjourney generative workflows in bulk.

It handles prompt queues, aspect ratios, style settings, and upscaling tasks. It keeps your generative staging pipeline completely optimized, running hundreds of queues in the background.

To enable all metadata scans, asset renaming, and automated generator pipelines in AutoStock, you link your connection keys inside the settings panel.

This puts you in full control of your resource usage. You can configure:

• OpenAI API Key: Powering our multi-modal vision models for instant asset analysis.
• Generation Tokens: For direct connection with Midjourney, Recraft, and Ideogram.
• Supabase credentials: To synchronize user settings, logs, and account metadata with your private database secure state.