microAI focuses on memory, latency, and older-device compatibility.
Compressed memory, response latency, context controls, and CPU performance settings are now part of the private AI direction.
LoomJ is XG's independent app studio. We build private AI, health, photography, file, and creator tools with a calm interface and a bias for ownership.
A growing ecosystem spanning local AI, food and fasting, photo metadata, compression, training, and creator portfolios.
Fully offline private AI with local models, semantic memory, and device-aware controls.
AI food scanning, calorie tracking, fasting rhythm, and iCloud sync in a playful notebook.
AI metadata generation for stock photographers, video creators, and visual archives.
Compression, archiving, encryption, and EXIF privacy cleaning for everyday sharing.
Interval timing, voice guidance, punch-power testing, and training history for fighters.
Motion, medical illustration, 3D visual systems, and creator-grade stock footage.
Across LoomJ, the recurring idea is simple: apps should respect your data, reduce friction, and make small workflows feel more deliberate. The tools can be advanced. The experience should stay quiet.
Local models, local memory, and on-device conversations for personal AI workflows.
Compress, archive, clean metadata, and share without exposing more than needed.
Metadata, search, and stock submission workflows for people who make visual work.
Health and training tools built around gentle tracking, not permanent noise.
LoomJ is not a category. It is a way of making tools: focused, durable, privacy-aware, and quietly useful.
We treat user data, attention, and device limits as product constraints.
Simple to use, unobtrusive by design, powerful when needed.
Independent tools for people who want to stay independent.
Careful interfaces, thoughtful defaults, and continuous improvement.
The product work and stock portfolio share one direction: tactile minimalism, utility, and visual clarity.
fotometa exists because upload metadata is repetitive, but search visibility matters. It turns creator logistics into a manageable system.
Open fotometaThe 4x-image portfolio gathers visual experiments and stock footage across design, technology, and medical illustration.
View iStock portfolioA compact log of the current product direction. More formal release notes can come later.
Compressed memory, response latency, context controls, and CPU performance settings are now part of the private AI direction.
Compression performance improved, with EXIF cleaning and archive workflows positioned as everyday file hygiene.
Health tracking, creator metadata, and training tools round out LoomJ's practical app family.