Taskling

Inspiration

Everyone in our team has at some point added some sort of gamification aspect to their tasks. This led us to think of a good LLM assistant to take care of the gamification aspects. This project automates personal organization by gamifying task management. Users can input tasks, which are automatically analyzed for complexity and then they are broken into subtasks and assigned experience points. The system leverages multiple agents built using Google’s Agent Development Kit (ADK) to handle the various functionalities.

What it does

Given a brief or detailed description of a task, the agent will split it into chunks, assign a experience value to it and schedule the different chunks with appropriate timelines. It also provides rewards based on the tasks completed and keeps track of the scores.

How we built it

Tech Stack

Backend: Google ADK (multi-agent system), Flask (custom Calendar API) ⁠Frontend: Streamlit ⁠Integrations: Google Calendar API, Google Developer Cloud ⁠Hosting: AWS

User Workflow

• We used a Streamlit frontend and a Python backend to interface the LLM backend.
• Once user input is acquired, we pass it on to backend with appropriate prompt modifications.
• The backend decides the split of tasks and the required timeline for each sub task and schedules required modules
• The frontend then displays the decision and the output from the model

Challenges we ran into

Some segments of the api development were hard to get running. Getting appropriate LLM outputs and then formatting it was a challenge

Accomplishments that we're proud of

We were able to use the app for ourselves, and having built something we can use is something we are very proud of.

What we learned

• Start early
• Handle AI hallucinations

What's next for Taskling

We want to extend this with more features and integrations and make it the one stop agent for all sorts of tasks and gamification options.

Built With

• amazon-web-services
• python
• streamlit