GoalGrinder

Inspiration

Everyone wants to learn something new - whether it's baking, an instrument, or even a programming language. The problem is that we’re often just too busy with our current schedules to figure out when and how to fit learning into our lives. Traditional learning platforms ignore your actual availability, while calendar apps don't understand learning progression. With this in mind, we wanted to create a platform that can both manage all of your goals and use AI as a mentor that creates realistic and adaptive learning plans to fit your schedule.

What it does

GoalGrinder is a platform that manages all of your goals while also turning them into achievable, adaptable, calendar-integrated action plans through optional voice conversation.

Simply type or speak your goal: "I want to get better at leetcode for my interview in 3 weeks" and GoalGrinder will:

Analyze your Google Calendar to find available time slots Create structured learning sessions with progressive skill-building Schedule specific tasks with specific resources directly into your calendar Adapt in real-time when time conflicts arise Create a chatbot history for each new goal

How we built it

LLM-Based Multi-Agent System powered by Claude Sonnet 4 and LangGraph’s ReAct Agent framework:
Scheduling Agent: Breaks down goals into a schedule and goal plan
Rescheduling Agent: Redesigns the schedule and goal plan based on the user’s conflicts
Event Creator Agent: Translates and adds all plans into Google Calendar events
Google Calendar API tools for agents to use
Seamless and smooth web app to manage all tasks
Using Groq Whisper for ultra-fast speech-to-text processing and a Voice-First Experience:
Natural conversation that understands context and adapts lesson plan as you go

Challenges we ran into

Google Calendar API authentication issues
Multi-agent coordination while maintaining conversation context
Calendar API complexity handling all-day events and scheduling conflicts
Voice processing latency - balancing speed with accuracy

Accomplishments that we're proud of

Real Google Calendar integration - not fake demo data
Truly conversational AI
Intelligent adaptation - understands learning progression and reschedules smartly
Multi-agent architecture that actually works

What we learned

Planning with design documents for the team is far more crucial than actual code-writing
Taking a step back when stuck and consulting for new ideas will help clear confusions up
Understanding a good balance between abstracting functionality away with AI and coding