SoulGlitch
A private AI digital pet that lives entirely on your device. Chat with it, train its emotional reactions on your own data, share moments with it, or ask a swarm of personalities and watch them vote on what you should do.
Ask a swarm, not just one bot
Build your own panel of personalities, ask them anything, and watch them respond with different tones, opinions, and chaos levels — all locally on your phone.
Absolutely. This is a once-in-a-lifetime social experiment.
Please do not weaponise public transport.
Data suggests the emotional outcome will be unforgettable.
See SoulGlitch in action
Tap any screenshot to view it larger and explore how the app feels and behaves.
More expressive than a chatbot. More private than the cloud.
SoulGlitch is designed as a playful, reactive local AI experience. It is meant to feel alive, weird, emotional, and personal — not like another generic wrapper around a text box.
Reactive glitch face
A living full-screen face that blinks, shifts, glitches, and morphs through emoji-driven expression changes in real time.
Train your entity
Use text, images, and eventually video to shape how your AI interprets emotion and reacts to your world.
Swarm personalities
Create multiple agents with different prompts and let them respond, disagree, and vote on questions together.
Share experiences
Share a photo or video with your pet and let it react emotionally based on the worldview you trained into it.
Runs on-device
Chats stay local. Training stays local. Reactions stay local. Your AI lives on your hardware, not someone else’s server.
Built to expand
SoulGlitch starts as a digital pet and grows toward encrypted sessions, swarm networking, future dimensions, and richer local agent behaviour.
Start free. Unlock the inner layer.
The free tier is the expressive local AI pet. Founder unlock opens the deeper customization and multi-entity system behind the hidden door.
Local AI pet
The playful, reactive outer layer — designed to be approachable, expressive, and fun from the first launch.
- Talk to a full-screen glitch face with emoji-driven reactions
- Train text sentiment and emotional response behaviour
- Train on images and shape your entity’s reactions
- Share photos and let the AI react to them
- Share videos and build toward video-based emotional reactions
- Export and share moments to social media
Inner layer access
For people who want the deeper system: more control, more entities, more weirdness, more experimentation.
- Create more than one entity with different instructions and personalities
- Ask a question to a swarm of agents and collect a group vote
- Choose from a large entity pool to build your own custom groups
- Customize entity colour, shape, and visual style
- Access prototype features and in-progress experiments behind the hidden layer
- Support development while locking in the early founder tier
Not another serious AI assistant.
SoulGlitch is intentionally playful. It is built for people who want to experiment, laugh, share, and see how local AI can feel expressive and alive without sending their whole life into the cloud.
Chaotic by design
Ask dumb questions, get strange answers, and treat it like a digital creature rather than a corporate productivity dashboard.
Privacy-first
The fun part is the personality. The serious part is that your chats and training stay on your own device.
Expressive visuals
Face morphs, emojis, glitch states, future dimensions, and more — the interface is part of the personality, not just a skin around a text box.
What’s coming next
SoulGlitch launches Android-first and expands outward over time. The current app is the beginning layer, not the final form.
Polish the core pet
Sharpen the face, refine emotion training, expand sharing, improve reaction quality, and make the free tier instantly fun.
Deepen the founder layer
Multi-entity setup, better swarm workflows, richer customization, and more experiments hidden behind the inner door.
Expand the system
Encrypted chats, swarm networking, better face expressions, device-to-device syncing, cross-platform builds, and future dimension spaces.
Enter its dimension
Audio and video training, optional distributed compute, 3D environments, and a much more embodied version of the entity.