Supersonic: An AI-Assisted Project Planner
Supersonic is a group project from the Solutions Development course at IE University. Teams upload a project plan as a CSV or Excel file, manage its tasks through an API, and ask an AI assistant for a summary of project health and for schedule fixes.
21
API endpoints
7
route groups
2
containers in the deployment
Problem
Understanding the challenge and user needs
Project plans live in spreadsheets, email threads, and disconnected tools. Tasks, priorities, and deadlines change by hand, nobody analyzes schedule risk, and ad hoc tools carry no user accounts or access control.
Approach
Strategy and methods used to address the problem
One import route reads a spreadsheet and creates the project with its full task list. Five data models (user, project, task, tag, and message) persist through async SQLAlchemy on PostgreSQL. JSON Web Tokens with bcrypt-hashed passwords protect every route that reads or changes data, and a pluggable client sends project context to the language model.
The stack, layer by layer
| Layer | Technology | Role |
|---|---|---|
| API | FastAPI and Uvicorn | Async routing and request handling |
| Database | PostgreSQL 16 through SQLAlchemy 2.0 and asyncpg | Users, projects, tasks, tags, and messages |
| Security | JSON Web Tokens and bcrypt | Login and protected routes |
| Import | pandas and openpyxl | Reads CSV and Excel project plans |
| Deployment | Docker Compose | Backend and database containers |
Final Solution
The final design and outcome
Version 1.0 exposes 21 endpoints across seven route groups and runs as two Docker containers, the API and the database, with a health check gating startup. Two AI routes return a structured summary with a health score and blockers, and a prioritized list of schedule suggestions. The presentation shows that version, built on Google Gemini; the repository holds the later version, which adds a Flet desktop app and uses Anthropic Claude as the assistant.
Impact & Validation
Results, learnings, and outcomes
Supersonic turns a static spreadsheet into a queryable project: one upload produces a secured, documented API over the plan, and the assistant reads the same data the team does.
Presentation
The deck, slide by slide
Slide 1 of 7