Products / shipped iOS apps / on-device direction

Real App Store products with icon, screenshots, and build notes.

This page is intentionally practical: each project includes the current App Store identity, screenshot set, product intent, and the core stack choices behind delivery.

Utilities / iOS

Easy PDF Pro

On-device PDF workflow assistant focused on privacy clarity, local processing, and predictable completion speed for everyday document tasks.

Core Stack Swift, native PDF pipelines, local inference surfaces, low-friction UX flow

On-device Model Path GGUF + llama.cpp is used for compact summarization and extraction helpers where latency and private context handling matter most.

Apple Foundation Models Layer For supported devices, Apple Foundation Models are used as a native completion layer for rewrite and structured output assistance.

Engineering Focus Token budget control, deterministic prompt templates, and fallback routing keep output stable for production utility workflows.

View on App Store Local case page

Lifestyle / iOS

CompaniPet

Private AI pet chat product with custom pet creation and low-pressure companion UX. Built around warm interaction while keeping boundaries explicit.

Core Stack Swift, on-device chat loop, role memory handling, companion safety framing

On-device Model Path GGUF runtime with llama.cpp drives personality-consistent chat turns for offline or low-connectivity sessions.

Apple Foundation Models Layer Apple Foundation Models are used for short-form rewrite and tone balancing to keep responses warm, concise, and safer by default.

Engineering Focus Session memory pruning, role-card constraints, and response guardrails keep long conversations coherent without overgrowth.

View on App Store Local case page

Productivity / iOS

Crush On

AI conversation practice app for social confidence and private interaction flow. The product direction emphasizes supportive scenarios and on-device core chat experience.

Core Stack Swift, interactive chat state, local-first inference path, subscription surface

On-device Model Path Small GGUF instruction models via llama.cpp are tuned for scenario rehearsal, quick turn-taking, and lower perceived delay.

Apple Foundation Models Layer Where available, Apple Foundation Models support natural phrasing and concise summary reflections after simulated chats.

Engineering Focus Prompt contracts for tone, response-length caps, and safety-first scenario tagging maintain consistency across personality modes.

View on App Store

Utilities / iOS

Should I Decide

On-device decision helper for relationship, career, and life-choice reflection. The product goal is clarity support, not replacing user judgment.

Core Stack Swift, private local analysis flow, structured decision response templates

On-device Model Path GGUF models with llama.cpp handle local reasoning outlines and multi-option comparison prompts without sending private context externally.

Apple Foundation Models Layer Apple Foundation Models support concise reframing and checklist synthesis for users who need fast, actionable next steps.

Engineering Focus Decision-tree prompt structures, confidence scoring language, and bounded answer format improve reliability and user trust.

View on App Store

Health & Fitness / iOS

AURA UP

AI beauty analysis product focused on facial scan interpretation, symmetry insight, and personalized glow-up protocol guidance.

Core Stack Swift, image analysis pipeline, personalized recommendation layers, progress tracking

On-device Model Path Local GGUF assistants are used for routine explanation, habit suggestions, and daily protocol guidance tied to scan output.

Apple Foundation Models Layer Apple Foundation Models provide concise rewrite and protocol summarization for clearer next-step communication on device.

Engineering Focus Image-to-text structuring, recommendation templating, and conservative health wording keep guidance actionable and compliant.

View on App Store

Entertainment / iOS

Star Secret

Anonymous confession and chat experience with gamified unlock flow. Built around intensity pacing, privacy-first interaction, and rapid loop engagement.

Core Stack Swift, anonymous chat session flow, coin economy UI, moderation boundary handling

On-device Model Path GGUF + llama.cpp powers short reveal expansions and conversation continuation while keeping private turn context local to the device.

Apple Foundation Models Layer Apple Foundation Models are used for tone normalization and concise clarification when chats require a safer narrative trajectory.

Engineering Focus Risk keyword throttling, reveal-level pacing rules, and moderation-oriented prompt guards reduce unsafe drift in anonymous sessions.

View on App Store