Inspiration
We were inspired by the limitations of traditional web extensions in providing on-the-fly, deep content analysis. Our goal was to leverage the new Chrome Built-in AI Challenge 2025 APIs—specifically Gemini Nano, the Prompt API, and the Summarizer API—to create a privacy-first tool that converts unstructured webpage data into actionable, visual intelligence, all processed locally on-device.
What it does
Interacto (AI Visualizer Pro) provides dual-mode, professional webpage analysis.
Screen Capture Mode: It extracts comprehensive data (text content, headings, links, metadata, performance metrics) from the current webpage.
Manual Input Mode: It acts as a ChatGPT-like interface for user-typed questions.
Analysis: It uses Gemini Nano to generate concise, structured summaries, extract key topics, assign importance scores (0-10 scale), and provide contextual insights.
Visualization: It dynamically generates Bar, Pie, and Line charts using Chart.js based on the AI analysis, with export functionality (PNG).
How we built it
We built Interacto as a Manifest V3 Chrome Extension. The core logic is in popup.js, handling the front-end interaction, mode switching, and history management. Data is extracted from the active tab using a content script (content.js) via chrome.scripting.executeScript(). This raw data is then structured into a detailed prompt for processing with LanguageModel.create() and session.prompt() (the Prompt API/Gemini Nano). Finally, the structured JSON response is parsed to update the chat interface and render interactive charts using Chart.js.
Challenges we ran into
A major challenge was ensuring the stability and performance of the AI pipeline, particularly managing the latency associated with the initial Gemini Nano download and availability check. We had to implement robust error handling and a real-time availability checker to guide the user. Parsing the complex JSON output from the AI response also required careful handling and fallback logic to ensure the UI remained functional even with minor parsing errors. We also faced size limits, optimizing the extraction to a balance of 8,000-12,000 characters for optimal AI performance.
Accomplishments that we're proud of
We are most proud of achieving full compliance with all Chrome AI Challenge 2025 requirements, including the integration of the Prompt and Summarizer APIs, and running on Manifest V3. The implementation of the dual-mode operation (Screen Capture and Manual Input) provides a highly versatile user experience. The dynamic visualization feature, which generates interactive charts and allows PNG export, successfully transforms text-based analysis into visual data.
What we learned
We gained deep knowledge of the Chrome Built-in AI architecture and the nuances of the new Prompt API lifecycle, particularly session management and efficient data transmission. We learned how to effectively scope system prompts to guarantee structured JSON output from a large language model. Furthermore, we refined our skills in building polished, responsive user interfaces for Chrome extensions using modern techniques like Glassmorphism, CSS animations, and Chart.js integration.
What's next for Interacto
Our future roadmap includes:
Phase 1: Multi-language support for analysis and custom system prompt configuration.
Phase 2: Advanced features such as PDF report generation and CSV data export.
Phase 3: Exploring platform expansion to other browsers and mobile support.
Log in or sign up for Devpost to join the conversation.