NanoPost

Inspiration

In today’s fast-paced social media world, content creators, marketers, and small businesses often struggle to generate engaging posts quickly. We wanted to build a tool that could transform a simple idea into high-quality social media content instantly, while keeping user data private and accessible offline. The inspiration came from seeing how many people spend hours crafting tweets, Instagram captions, or short video scripts — and realizing AI could automate that process efficiently, locally, and safely.

What it does

Through this project, we learned: How to integrate Chrome Built-in AI APIs, especially Prompt API, Rewriter API, Proofreader API, and Translator API, into a web application. The challenges of handling multi-modal input (text, images, audio) for content generation. Techniques for offline-first AI applications, including caching AI models and processing requests locally with Gemini Nano. Best practices for designing user-friendly UI/UX for content creators while respecting privacy.

How we built it

Frontend: Built with React for a responsive web interface. Users can input a topic, upload an image, or even record a short voice note as inspiration. AI Integration: Prompt API: Generates the initial draft of tweets, Instagram captions, or YouTube Shorts descriptions. Rewriter API: Provides alternative wording and tone options. Proofreader API: Ensures the text is grammatically correct. Translator API: Translates posts into multiple languages for global reach. Local Processing: By leveraging Gemini Nano, all AI computations run client-side, ensuring privacy, low latency, and offline capability. Output: Users can preview, edit, and export content directly to their favorite social media platforms.

Challenges we ran into

Balancing creativity and control: AI sometimes generates off-topic content. We solved this by iteratively refining prompt templates and adding user feedback loops. Offline model performance: Running AI locally required careful memory and computation management to keep response times under 2 seconds. Multi-platform content formatting: Different social networks have different length, hashtag, and style requirements. We implemented post-formatting rules to adapt AI output automatically.