HackMate

Framing the Problem

Hackathons are fast-paced environments where teams are expected to brainstorm, plan, build, and pitch—all within a day or two. Given the rush, many teams face the same pain points:

• Wasting time on idea discussions

• Poorly distributed roles

• No clear sprint planning

• And finally, scrambling to write a pitch just before deadline

This especially affects early-career developers or mixed-experience teams who lack a clear structure. We asked ourselves: What if teams had a digital teammate guiding them through the entire hackathon lifecycle?

Idea Explanation

We built HackMate—an AI-powered hackathon co-pilot.
HackMate helps teams go from concept to pitch, faster and more confidently. It guides users through:

• Idea Generation – Powered by Gemini AI, it turns rough prompts into clear, actionable project ideas

• Team Dynamics – Assign roles and responsibilities to keep everyone aligned

• Sprint Planning – Plan milestones with deadlines and visualize progress with a live tracker

• Pitch Coach – Automatically generate pitch drafts or write them collaboratively

Instead of juggling Notion docs, spreadsheets, and whiteboards, teams can now manage their hackathon journey from a single, intuitive interface.

Implementation

HackMate is a multi-page web app built using the Dash framework (by Plotly) with dash-bootstrap-components for responsive UI. Each page (Idea Generator, Sprint Planner, Team Hub, etc.) is modular and controlled via dedicated callback files.

• Frontend: Dash handles dynamic component rendering and routing

• AI Integration: Google Gemini 1.5 Pro API is used for idea and pitch generation

• State Management: We used dcc.Store for session persistence across pages

• Auth & DB: Simple authentication is backed by SQLite with hashed passwords

All data—from sprint plans to team roles—is exportable with one click.

Challenges

Managing state and user sessions across multiple Dash pages was a major challenge. Since Dash is essentially a single-page app under the hood, getting login persistence to work smoothly (especially with things like navbar changes and protected pages) required careful use of dcc.Store and custom triggers.

Also, Gemini AI doesn’t return structured data, so crafting prompts that consistently output usable pitch sections (like Problem, Solution, UVP) took a lot of trial and error.

Accomplishments

We’re proud that HackMate isn’t just a demo—it’s a complete tool:

✅ AI-generated ideas that feel usable
✅ Sprint planner with milestone editing, deletion, and progress tracking
✅ Dynamic team role assignment and export
✅ A pitch-writing coach powered by Gemini, with both manual and AI-assisted modes
✅ A clean, intuitive UI with a persistent session-based login system

And it all works in one place. No external tabs. No mess.

Next Steps

• Real-time team collaboration (shared session editing)

• Hosting and deployment for public use

• GitHub integration to turn sprint plans into issues