LlamaBox brand kit.
Rules for creating clear, accurate LlamaBox content. This is a production reference, not a media asset download.
Brand promise
Private offline AI chat for Android. Compatible GGUF models run on the phone through llama.cpp; local chat needs no LlamaBox account or LlamaBox server.
Voice
- Direct and grounded: explain the mechanism, not just the benefit.
- Calmly technical: plain language first; name GGUF and llama.cpp when it adds useful precision.
- Honest about limits: CPU is the default execution path; model fit and speed depend on the device.
Approved message pillars
- Private by architecture: inference stays on the device.
- Useful offline: chat works after a compatible model is downloaded or imported.
- Built for Android: Android 7.0+ arm64 devices; CPU-default inference in the current build.
- Model choice: compatible GGUF models, with fit determined by available memory and device limits.
Visual system
- Base: near-black
#07070B; elevated surface#12121A. - Accents: violet
#A78BFAand cyan#22D3EE. - Type: Inter for UI/body; Cormorant Garamond for editorial display moments.
- Form: generous dark space, translucent layers, rounded geometry and one clear phone/device focal point.
Logo downloads
4K PNG exports for approved brand use. Use the matching background version; do not crop, recolor, or add effects.
Claims guardrails
- Do not claim GPU acceleration, universal model compatibility, benchmark results, or availability beyond verified Google Play regions and devices.
- Do not imply that downloading models is offline; only local inference/chat is offline after the model is present.
- Do not use “anonymous” as a substitute for the stated no-account, no-LlamaBox-server boundary.
Required carousel footer
Every exported carousel slide must include LlamaBox · llamabox-ai.github.io/brand-kit.html in its footer safe area. The link is part of the asset specification and must remain legible after platform compression.
Content source
The first unrendered carousel brief and its validation contract live in the repository at assets/carousels/ideas/001-private-by-architecture.json. Create or render assets only after the copy and any product claims have been approved.