Check out the project on Devpost and view the source code on GitHub.
I spent a weekend at HackNC building something to help manage all of my other side projects. What started as a simple “Wouldn’t it be cool if an AI actually rang you to ask if you’d done your work?” turned into AccountaBuddy, a lightweight app that does more than fire off push notifications - it actually calls you, celebrates your wins, and helps you problem-solve when tasks stall.
The Why: When Reminders Aren’t Enough
I’m not the only one who’s ever ignored yet another notification. A to-do list can feel like a list of obligations, or worse, just noise. AccountaBuddy was born from the idea that human accountability is powerful - so why not let technology step in as a friendly collaborator? Instead of sending one-way reminders, it reaches out, listens to you, and adapts to your situation.
What It Does, in Practice
- Task creation & scheduling: Sign up, add your tasks, and pick check-in times.
- Real-time voice calls: At each scheduled moment, AccountaBuddy actually rings your phone via bland.ai.
- Dynamic follow-up: If you’ve completed a task, it cheers you on. If not, it asks what’s blocking you and suggests next steps.
- Interactive planning: You can reprioritize or break a big goal into bite-sized actions in an on-the-fly chat.
Rather than a static list of “done” or “not done,” it turns to-dos into a conversation. The UI was inspired by our code editor, Cursor, with the AI proposing modifications to the to-do list and the user confirming or denying them.
Under the Hood
Despite having just two developers and 24 hours, we got a working proof of concept by combining:
- Next.js + React for the UI and routing
- Prisma for defining and managing users and tasks in our database
- bland.ai for dialing out and handling call flows
- ChatGPT to guide the conversation, interpret responses, and offer personalized suggestions
What We Struggled With
- State management: React’s component lifecycle can be tricky when you need live updates for tasks that might change status mid-call.
- API choreography: We had to make sure calls, chat logic, and database updates stayed in sync without visible delays.
- Background jobs: Scheduling outbound calls reliably—especially under a tight timeline—meant building a small worker queue and retry logic.
Takeaways & Next Steps
By the end of the hackathon, AccountaBuddy was already helping us plan the remaining tasks, but there’s still more we’d like to do in the future:
- Stripe integration for a lightweight subscription model
- Live voice conversations powered by ChatGPT Advanced Voice Mode, so you can brainstorm out loud with your AI buddy in real time
- Deeper analytics, to show you trends in your productivity rather than just individual tasks