Getting started
1. Choose your AI engine
StockMeta AI can generate metadata four ways — two local, two cloud:
- Local (private): install Ollama or LM Studio — both free — then open the Models tab and download a recommended Gemma vision model. Everything runs offline on your Mac. Because the App Store build is sandboxed, start Ollama yourself first with
ollama serve. - Cloud: open Settings ▸ AI Backend and choose Google Gemini or ChatGPT / OpenAI, then paste your own API key. Requests go straight from your Mac to that provider — never through a server of ours.
2. Add your files
Drag photos, illustrations, or videos onto the analyzer, or click to browse. You can add a single file or a whole batch at once.
3. Generate & review
Click Generate Metadata (or press ⇧↩). Each row fills in live with a title, a one-sentence description, and 15–49 relevance-ranked keywords. Edit any field inline before exporting, or regenerate a single row after adding a note to it.
4. Export
Use Export to save an Adobe Stock CSV, a Shutterstock CSV, a generic keyword CSV, or XMP sidecar files that embed metadata directly into your images.
Choosing a local model
The Models tab recommends a size based on your Mac's memory — Gemma 3 4B for 8 GB, Gemma 12B for 16 GB, Gemma 3 27B or Gemma 4 31B for 32 GB and up. Larger models are more accurate but need more memory and time. If results feel thin or generation is slow, try a different size. The app shows the approximate download size for each one.
Troubleshooting
The local engine won't start
Make sure Ollama or LM Studio is installed and running. The App Store build is sandboxed, so it can't launch Ollama for you — open Terminal and run ollama serve, or start the LM Studio server, then hit refresh in the Models tab. A model must be downloaded before generation will work.
A cloud backend returns an error or rate limit
Confirm your API key is pasted correctly in Settings ▸ AI Backend. Free-tier Gemini and OpenAI keys have rate limits — wait a minute, switch to a lighter model such as gemini-2.5-flash-lite, or use a local backend for large batches.
Keywords or titles look off
Add a note to the row (or a batch context that applies to every file) and regenerate — telling the model the location or intended use sharpens results a lot. A larger model gives richer detail. On the Max tier, the Rules tab lets you edit the underlying prompt for each backend and media type.
A purchase isn't recognized
Open Settings ▸ Plan and choose Restore Purchases, and make sure you're signed into the same Apple ID used to buy. Purchases are family-shareable.
Refunds
All purchases and in-app purchases are processed by Apple, so refunds are handled by Apple — not by us. To request one, visit reportaproblem.apple.com, sign in with your Apple ID, find StockMeta AI in your purchase history, and choose Request a refund. Apple reviews each request according to its refund policy. If you ran into a problem with the app, we'd love the chance to help first — email us before requesting a refund and we'll do our best to make it right.
System requirements
- macOS 14.0 (Sonoma) or later
- Apple Silicon recommended for local AI; Intel Macs work with the cloud backends
- For local AI: Ollama or LM Studio, plus enough free disk space and memory for your chosen model (sizes shown in-app)
- For cloud AI: your own Google Gemini or OpenAI API key and an internet connection
Terms of use
StockMeta AI is provided as a tool to assist with creating stock metadata; you are responsible for reviewing all generated titles, descriptions, and keywords before submitting them to any agency, and for complying with each agency's content and metadata policies. The app is provided "as is" without warranties of any kind to the extent permitted by law. Purchases are handled by Apple and governed by the Apple Media Services Terms. See our Privacy Policy for how data is handled.
Still need help?
Email buyanskyyd@gmail.com with your macOS version, which AI backend you're using, and a short description of the problem. Screenshots help a lot. You can also use Settings ▸ Feedback ▸ Report a Bug in the app, which pre-fills a draft with your version and backend already filled in.