MinuteMind

AI-powered meeting summarization agent that turns raw meeting notes into structured, actionable insights using Amazon Bedrock.

Comment

🧠 MinuteMind – AI Meeting Summarizer

MinuteMind is an AI-powered meeting summarization tool built for the AWS AI Agent Global Hackathon. It helps users turn messy, unstructured meeting transcripts into clear, structured, and actionable summaries in just seconds — saving time and boosting team productivity.

🚀 Overview

Meetings are essential for collaboration, but they often leave behind long, unstructured notes that are hard to review. MinuteMind solves this problem by automatically generating structured summaries from raw meeting transcripts, including:

📝 Topic – the main focus of the meeting

✅ Decisions – key outcomes and agreements

📌 Action Items – tasks and next steps

💡 Next Steps – suggestions and follow-ups

Simply paste your meeting notes, and MinuteMind transforms them into a concise, human-readable summary instantly.

🌟 Inspiration

In today’s fast-paced work environment, employees attend countless meetings filled with lengthy discussions and complex decisions. Often, people lose focus, miss key points, or struggle to recall important details afterward.

We wanted to solve this common frustration and help teams instantly understand what was discussed, what decisions were made, and what actions need to be taken — without replaying or rereading the entire meeting.

🤖 What It Does

MinuteMind transforms raw meeting transcripts into clear, structured summaries within seconds. It automatically identifies and organizes key elements such as Topics, Decisions, Next Steps, and Action Items, helping users quickly grasp essential takeaways.

Whether someone missed the meeting, zoned out halfway, or just wants a quick recap, MinuteMind ensures they stay aligned with the team and project goals.

🛠️ How We Built It

We built MinuteMind with a React frontend for a clean, responsive user interface and an Express.js backend to handle API requests.

Here’s how it works:

Paste any meeting transcript into the web app.

The backend sends the text to the Amazon Titan Text Express model via AWS Bedrock.

The LLM analyzes the content and generates a structured summary.

The result is displayed instantly on the frontend, organized into easy-to-read sections.

We used AWS Bedrock AgentCore for orchestrating the agent workflow and AWS IAM for secure access control. All environment variables and API keys are managed securely with .env.

🧗‍♂️ Challenges We Ran Into

🧠 Prompt engineering & summarization accuracy: Designing prompts that consistently generate structured summaries across different meeting styles was challenging.

⏱️ Time constraints: Building and refining the system end-to-end in a short hackathon period required rapid iteration and prioritization.

🧩 Structured output formatting: Ensuring that the model consistently produced summaries with sections like Topic, Decisions, Next Steps, Action Items required extensive testing and tuning.

🏆 Accomplishments We’re Proud Of

✅ Built a working prototype that summarizes lengthy meeting transcripts into clear, structured summaries in seconds.

📊 Implemented structured output generation that organizes information into actionable categories.

🧩 Created a simple and intuitive UX that requires no technical background — just paste your transcript and get insights instantly.

☁️ Integrated AWS Bedrock and Amazon Titan seamlessly to build a fully functional end-to-end AI agent.

📚 What We Learned

We learned how crucial prompt design and iteration are when working with LLMs — small wording changes can drastically affect output quality. We also gained valuable experience in integrating frontend and backend components and saw how user experience design plays a key role in making AI-powered tools practical and valuable in real-world scenarios.

🧪 Tech Stack

Frontend:

⚛️ React – for building the user interface

🎨 CSS – for clean and responsive styling

Backend:

🟢 Express.js – RESTful API server

🌐 CORS & dotenv – environment configuration and secure API calls

AI & Cloud:

☁️ AWS Bedrock AgentCore – for agent orchestration

🧠 Amazon Titan Text Express – large language model for summarization

🔑 AWS IAM – secure access control and key management

Other:

📁 GitHub – version control and collaboration

🌐 Fetch API – frontend-backend communication

✨ Key Features

📄 Instant AI Summaries: Paste raw meeting text and receive a structured summary in seconds.

⚙️ Full-Stack Integration: Smooth connection between frontend (React) and backend (Express).

🔐 Secure Environment: Secrets managed via .env and never exposed.

☁️ Serverless LLM on AWS: Uses Amazon Titan via Bedrock — no model hosting required.

📱 Clean UI/UX: Simple and intuitive design for effortless summarization.

📦 Project Structure fullstack/ ├── client/ # React frontend │ ├── src/ │ │ ├── MainPage.js # Main UI page │ │ └── Summary.js # Summary display component ├── server/ # Express backend │ ├── index.js # API routes and Bedrock integration │ └── .env # (not committed) AWS keys & secrets └── package.json

🚀 What’s Next for MinuteMind

We plan to expand MinuteMind beyond transcript input by:

✍️ Multi-language support – English, Japanese, and Chinese

📁 File & API imports – directly ingest transcripts from various sources

🔔 Slack & Email integrations – deliver summaries automatically to where teams work

🤖 Agent-style follow-ups – e.g., “Generate next meeting agenda”

📊 Dashboard view – track decisions and follow-ups across multiple meetings and projects

✅ Why This Matters: Teams spend countless hours in meetings, but disorganized notes waste even more time afterward. MinuteMind eliminates this friction by delivering structured, actionable summaries with zero manual effort, making follow-ups easier, decision tracking clearer, and collaboration smoother — especially for remote and hybrid teams.

Share this project:

Updates