PROcrastinate

Logo

Inspiration

Like many groups at the start of this Hackathon, we were stuck for what to do. We brainstormed for hours and doom-scrolled to procrastinate. That all changed when we simply went for a walk. We realised that maybe procrastination wasn’t the problem, but specifically bad procrastination was. So we came up with PROcrastinate to help you put the “pro” in procrastinate.

What it does

PROcrastinate is here to help you procrastinate efficiently. It tracks your activity and detects when you’ve drifted off-task or hit a mental block (or even when you’ve worked too hard). It suggests smarter breaks to help you come back more focused. It adapts to how you’re working, encouraging you to take guilt-free breaks, and helps avoid wasting time.

How we built it

We were able to split our solution into four main subsystems:

A React desktop UI for onboarding, sessions, stats, break library, and settings
An Electron main process that owns windows, persistence, overlay behavior, IPC, and AI calls
a Chrome extension that captures live tab/browser telemetry
a trained OpenAI-backed reasoner that judges relevance between the current context and the session task

Challenges we ran into

One of the main problems we had was the behaviour detection. Initially we tried to specify too many specific “bad habits”, but this became too complex. We solved this by reducing bad habits to just two: deviation and inactivity. Another challenge was balancing helpfulness with annoyance, making sure the app intervenes at the right time without being intrusive. We also had some trouble with the chrome tabs, having to make sure the system determined the correct site the user is on and the AI correctly categorising it as good or not.

Accomplishments that we're proud of

We were proud of:

Creating a system that uses ML to analyse whether you are productive or not
Making an aesthetic UI
Designing a unique angle on productivity that people instantly relate to
Finishing this task with only 3 people

What we learned

We learned that less complexity often leads to better products. By focusing on core user states instead of many edge cases, we made the system more robust and easier to build. We also learned how important tone and UX are, the same feature can feel helpful or annoying depending on how it’s presented.

What's next for PROcrastinate

In the future, we would like to:

Implement an insights page, giving users the ability to see their productivity vs procrastination time
Settings page to customise colour palette, notifications, etc.
Improve detection using more behavioural signals
Incorporate a mobile companion app
Letting the user rate their breaks to maximise break efficiency over time
Personality of AI response