TinyLLM Labs raises $18M to launch a 3MB on-device UX assistant

AI · 3 min read

Investors including Harbor Lane and Founders Collective backed the financing, which the startup will use to expand hardware partnerships and deliver optimized runtimes for Android and embedded Linux. TinyLLM Labs also plans to open a limited developer program for OEMs.

PocketMuse targets mobile UI augmentation tasks like contextual suggestions, microcopy generation and voice-first shortcuts, while running fully offline to avoid latency and privacy issues. The company says the model uses a distilled architecture with aggressive quantization and custom tokenization for UI text.

Early trials with two handset makers showed sub-200ms median latency on mid-range processors and no network dependency for chat or assistive features. TinyLLM Labs will release an SDK and developer guidelines in beta next quarter.