Photocore
Local AI photo search, even for disconnected drives.
Hey everyone, I wanted to share the story and philosophy behind my latest project, PhotoCore.
Like many developers and digital creators, I have accumulated terabytes of photos over the years. To keep our internal Mac storage clean, the standard move is to offload these archives to external SSDs, hard drives, or NAS units. The real issue starts when those drives go into a drawer. Those memories become completely invisible. If you need to find a specific shot from a few years ago, you have to plug the drives in one by one, wait for them to spin up, and guess where it is. The commercial cloud alternatives solve this, but they impose a perpetual monthly subscription tax and require you to completely surrender your data privacy and control to big tech monopolies.
I built PhotoCore to break this cycle. I wanted a tool that lets us become the absolute owners of our personal information again, pushing the incredible local processing power of Apple Silicon chips to their absolute limits.
PhotoCore is a native macOS application that brings AI semantic search directly to your machine, without a single byte of data ever leaving your computer. The core architecture allows you to do something unique: once the initial scan is done via CoreML, the app stores the embedding vectors and optimized previews locally. This means you can search, filter, and browse through your entire photo archive even when your external hard drives or NAS are physically disconnected and sitting turned off in a drawer.
To move beyond simple keyword search, I designed two specific features for semantic exploration and gallery organization:
Semantic Bridge: This allows you to select any two images and uncovers the visual narrative showing the conceptual connection and story that links them.
Constellation: This intelligently organizes your gallery, automatically grouping files based on their semantic context to bring out key life moments like weddings or travel memories.
From an indie hacking and bootstrapping perspective, the most rewarding part has been proving that you can build advanced software while rejecting cloud servers and monthly fees. Because all the heavy processing runs entirely on the client hardware, it eliminates recurring infrastructure overhead for my business while offering a clean, one-time purchase model with no subscriptions for the users.
I would love to get your thoughts on this privacy-focused, local-first approach. How do you currently manage your offline digital archives?
1 Comment
The part I found most interesting isn't the AI search—it's decoupling discovery from storage. Most archive systems assume the drive has to be online before the photos become useful again. Making an offline archive searchable changes it from "cold storage" back into something you can actually use without friction.