🤖 AI Summary
This work addresses key challenges in deploying small language models (SLMs) on mobile devices—including erroneous outputs, constraint violations, contextual degradation, latency incompatibility, and unstable model selection—by integrating Gemma-2.6B and Qwen3-0.6B into the Android word game Palabrita. The authors simplify the LLM’s role from generating full puzzles to producing only three hints and introduce a deterministic fallback mechanism. Embracing a “less is more” principle, they derive eight design heuristics for SLM integration on mobile platforms and systematically categorize five failure modes along with corresponding mitigation strategies. Through techniques such as multi-layer defensive parsing, context-aware retries, conversational turn rotation, and responsibility minimization, the approach enables stable, production-grade SLM deployment, substantially improving system reliability and user experience.
📝 Abstract
On-device Small Language Models (SLMs) promise fully offline, private AI experiences for mobile users (no cloud dependency, no data leaving the device). But is this promise achievable in practice? This paper presents a longitudinal practitioner case study documenting the engineering challenges of integrating SLMs (Gemma 4 E2B, 2.6B parameters; Qwen3 0.6B, 600M parameters) into Palabrita, a production Android word-guessing game. Over a 5-day development sprint comprising 204 commits (~90 directly AI-related), the system underwent a radical transformation: from an ambitious design where the LLM generated complete structured puzzles (word, category, difficulty, and five hints as JSON) to a pragmatic architecture where curated word lists provide the words and the LLM generates only three short hints, with a deterministic fallback if it fails. We identify five categories of failures specific to on-device SLM integration: output format violations, constraint violations, context quality degradation, latency incompatibility, and model selection instability. For each failure category, we document the observed symptoms, root causes, and the prompt engineering and architectural strategies that effectively mitigated them, including multi-layer defensive parsing, contextual retry with failure feedback, session rotation, progressive prompt hardening, and systematic responsibility reduction. Our findings demonstrate that on-device SLMs are viable for production mobile applications, but only when the developer accepts a fundamental constraint: the most reliable on-device LLM feature is one where the LLM does the least. We distill our experience into eight actionable design heuristics for practitioners integrating SLMs into mobile apps.