Group Sync

• 

What Inspired Us:

Group projects are supposed to be collaborative, but in reality they are often chaotic. Tasks are unclear, workloads become unbalanced, and deadlines sneak up because no one truly knows how far along the team is. We have all worked on teams where some people are overloaded while others do not know what to do, or where the project lacks structure until it is too late.

We wanted to fix this. We envisioned an AI powered project coordinator that removes confusion, enforces fairness, and helps student teams work together more efficiently. This sparked the creation of Group Sync, an adaptive system that manages a group project from start to finish.

How Group Sync Works

Group Sync is an adaptive, multi agent project management system built with LangGraph. It functions like an intelligent project manager that automates planning, task allocation, progress monitoring, and dynamic adjustments.

1. Understanding the Assignment

Students upload the assignment brief, marking criteria, and deadlines. A parsing agent extracts the key components, including: • required tasks • deliverables • milestones • dependencies • timing constraints

This turns vague text into a structured project blueprint.

1. Profiling Each Student’s Skills

Each student submits: • skills • past experience • availability • working style • CV or portfolio

The skill profiling agent converts this into competency vectors that quantify strengths and availability.

1. Breaking the Project Into Tasks

Using the assignment structure and deadlines, the task decomposition agent: • generates tasks and subtasks • identifies dependencies • estimates difficulty • predicts time requirements • maps everything relative to deadlines

This produces a complete task graph for the project.

1. Allocating Tasks to the Right People

The task allocation agent assigns tasks based on: • skill match • workload balance • time availability • urgency and difficulty

This gives each student a personalised task list that fits their strengths.

1. Monitoring Progress Continuously

As students work, a progress agent tracks: • expected progress • reported progress • delays or bottlenecks • workload imbalance

This allows Group Sync to maintain an up to date view of the project’s status.

1. Adapting in Real Time

If progress falls behind, Group Sync automatically: • reallocates tasks • adjusts deadlines • redistributes workload • provides feedback

Group Sync behaves like a real project manager that is always paying attention and responding quickly.

How We Built Group Sync

We used a LangGraph based architecture that orchestrates multiple agents working together. The system includes:

Assignment Parsing Agent

Extracts structure and requirements from unstructured text.

Skill Profiling Agent

Transforms user inputs into numerical competency vectors using embeddings.

Task Decomposition Agent

Breaks assignments into tasks with dependencies.

Task Allocation Agent

Matches tasks to students using scoring and reasoning.

Monitoring Agent

Tracks expected versus actual progress.

Reallocation Agent

Updates the plan when the project state changes.

Everything is connected within a shared LangGraph state that agents read and update as the project evolves.

What We Learned

We learned: • How powerful agentic workflows can be • The complexity of accurate task decomposition • How to translate subjective skills into structured data • The importance of strong state management • How difficult it is to model progress and compare expectations to reality

Challenges We Faced

1. Extracting Tasks From Messy Briefs

Assignment instructions vary widely. The model sometimes missed details or invented steps. Iterative prompting and validation helped solve this.

1. Fair Workload Balancing

Designing a scoring system that considered skills, difficulty, time, and workload required experimentation.

1. Coordinating Shared Project State

Ensuring that updates from one agent did not break another agent’s reasoning was a major challenge.

1. Creating Clear Visuals

We invested time refining visual diagrams and arrows to clearly illustrate the workflow.

1. Working Under Hackathon Constraints

Time pressure pushed us to simplify, prioritise, and collaborate efficiently.

Summary

Group Sync is an adaptive AI system that ingests a project brief, profiles team members, decomposes the work, assigns tasks intelligently, monitors progress, and adjusts the plan automatically. Inspired by our own frustrations with group work, we built a fully agentic workflow that functions like a real project manager and helps student teams stay organised, efficient, and aligned.

Built With

• fastapi
• langchain
• langgraph
• react