I shoot senior portraits for a living, which means I'm sitting on a Lightroom catalog with tens of thousands of images of other people's teenagers in it, and until recently, essentially none of them had a title, a caption, or any metadata beyond whatever my camera stamped on by default. Ask me to find "the senior girl in the burgundy dress from the fall shoot two years ago" and I'd be scrolling for twenty minutes, squinting at thumbnails, hoping I remembered the right month.
Every photographer says they'll go back and tag the catalog properly "when things slow down," a sentence photographers have been telling themselves since Lightroom 1.0, and things have never once slowed down. I said it for years. Then I did the math on how long it would actually take to hand-write captions and alt text for a few hundred thousand images and understood, with total clarity, that I was going to die first.
Luxii is the plugin I built instead of dying first. It's not a separate photo library or a "find any shot instantly" app living apart from your workflow, whatever the tagline on this card says. It's a Lightroom Classic plugin: select anywhere from one photo to a hundred, click once, and it writes real metadata, title, headline, caption, alt text, SEO-tuned keywords, nineteen IPTC fields total, straight into your existing catalog. No export. No reimport. No second piece of software with its own login and its own opinions about how you should organize your life.
What It Actually Does, Not What The Card Above Says
Under the hood it's built on Claude, Anthropic's AI, not some homebrew tagging model trained on a spreadsheet of adjectives and a prayer. Lightroom renders each frame with your develop edits and crop already applied before Luxii ever sees it, so the tagging reflects the photo you actually delivered, not the flat raw file sitting underneath it. It never touches develop settings, crop, flags, or star ratings. Those stay exactly as you left them.
There's no review dialog and no manual tag categories to configure, which is a deliberate tradeoff, not an oversight. The AI writes the fields, you don't approve each one individually, because if I made you review three hundred photos one at a time I'd have built a slower version of the exact problem I was trying to solve.
The part I actually cared about, more than the metadata itself: photos get analyzed in memory and thrown away immediately after. Nothing gets stored on a server somewhere, and nothing gets used to train anybody's model. I shoot portraits of minors every week. I was not about to ship a tool that phones a stranger's photo library home for safekeeping, mine or anyone else's.
What It Costs And Where It Runs
It's a thirty-nine dollar one-time license, not a subscription, because I'm not interested in renting you a plugin forever for something you'll use in bursts twice a year around graduation season and holiday minis. That includes five hundred image credits to start, the first twenty-five are free with no card required, and credit packs run fifteen dollars for a thousand images up to forty-five for five thousand, credits that don't expire because nothing about my own tagging backlog respected a monthly billing cycle either.
It runs on macOS today. Windows is finished and in final testing as of this writing. There's also a browser-based trial if you want to test it on a handful of JPEGs before installing anything, twenty-five images, no Lightroom required, no plugin to install just to see whether the metadata's actually any good.
Why This One Exists At All
I didn't build this to sell software. I built it because I was staring down a catalog that would take me longer to hand-tag than I have left on this earth, and once it worked for me, it seemed insulting to keep it to myself while every other portrait and event photographer I know has the exact same untagged pile sitting in the exact same catalog. Ask around at any photography meetup and you'll hear the same confession from three different people before the coffee's gone: a hard drive full of beautiful, completely unsearchable work.
If you've got fifty thousand images with nothing but a camera-generated filename to your name, start at luxii.app. Try the first twenty-five for free and see if the captions sound like something you'd actually write. Tag a hundred RAWs before your coffee's cold. Then go shoot something instead of typing captions.