Smart Meeting Assistant

Aspect Emphasized,TaglineFocus on Action,Meetings end. Actions begin.AI Integration,The meeting assistant that writes the to-do list for you.Productivity,"Turn talk into tasks, instantly."

Comment

Inspiration Frustration with the "Action Item Black Hole"—tasks lost between meeting notes and execution.

The need to efficiently transform discussion into structured, accountable tasks.

What it Does Provides a centralized, trackable list for all Action Items (Title, Owner, Due Date).

Uses AI (Gemini) to automatically extract structured data from raw meeting text.

Pre-fills the task creation form with AI-suggested details, minimizing manual data entry.

How it was Built Frontend: Simple HTML, CSS, and Vanilla JavaScript for the user interface.

Backend: Python Flask server to handle logic and securely host the AI key.

AI: Integrated with the Google Gemini API (google-genai SDK) for text analysis and JSON data extraction.

Security: API key is loaded via .env and accessed only on the server, not the browser.

Key Challenges CORS: Resolving cross-port communication security issues between the browser (port 5500) and Flask server (port 5000) using flask-cors.

LLM Reliability: Ensuring the Gemini model consistently returned pure, parsable JSON data through specific prompt engineering.

Key Learnings Mastered secure server-side API key management (Flask/dotenv).

Gained experience in prompt engineering for structured data extraction (JSON schema).

Debugged and implemented solutions for complex CORS security policies.

What's Next Integrate a persistent database (e.g., SQLite) to store action items permanently.

Extend AI capability to process full meeting transcripts (not just short descriptions).

Implement proactive calendar checks to verify owner availability for suggested deadlines.

Develop the Dashboard for team-wide productivity analytics.

Built With

Share this project:

Updates