Screen Shot

📝 GenAIHackathon Project Write-Up

🚀 Project Inspiration

AutoPlan was born out of a simple but universal frustration:

We often know what we need to do — but not when or how to get it done.

Whether it's students juggling coursework or professionals overwhelmed by meetings and deadlines, people rarely sit down to break tasks into manageable steps — let alone schedule those steps intelligently.

Our team wanted to build something that could:

Understand natural, messy input like “submit report by Friday midnight”
Recognize it as a task with a deadline
Decompose it into subtasks like “draft report” and “get feedback”
Return a structured, time-aware plan

We saw generative AI — specifically LLMs like Gemini — as the perfect assistant for this job. With just one sentence of input, AutoPlan helps users go from vague intention to executable plan.

🧰 Technology Stack

🔤 Languages

JavaScript (Node.js)
JSON (API responses and prompt formatting)

🧱 Frameworks & Libraries

Express.js – for building the backend server
@google-cloud/vertexai – Vertex AI SDK to interact with Gemini
CORS, body-parser, etc. – for development convenience

☁️ Platforms

Google Cloud Vertex AI – to power the Gemini 1.0 Pro LLM for planning
Localhost during development (http://localhost:3000)

🛠️ Tools

Git – for team collaboration

- VSCode – for development

📦 Product Summary

AutoPlan contains an intelligent planning backend that:

Accepts natural language input
Uses Gemini to extract and normalize events and tasks
Automatically generates subtasks for each item with clear deadlines

🧠 Features

Understands mixed-input natural language:
“I have a team meeting on March 23 and need to submit my report by the 24th.”
Parses and normalizes this into structured JSON:
- Events (with start/end time)
- Tasks (with deadlines)
Plans 2–3 prior steps per task/event with realistic due dates
Returns output in clean, frontend-friendly format

💡 Innovative Aspects

Modular pipeline:
understand → normalize → plan
Each powered by a targeted Gemini prompt
Flexible architecture for future extensions (calendar export, reminders, chat interfaces)

🎯 User Experience

Backend-first prototype
Ideal for future integration with calendar tools, chatbots, or productivity apps
Designed to act like a “mini project manager in your pocket” — without the overhead