NoStall

• 

  Home page allows you to add a task through voice or text

• 

  2min version page provides the quick sub task of the next listed task

• 

  calendar page shows the completed tasks of the particular date

• 

  Tasks page provides with the list of all tasks and also the 2 min version of the main task

Inspiration

There have been countless instances when we have a lot of work to do but end up doing nothing instead. This is because the very fact of doing a hard task tires us out than the task itself and we are not sure how to prioritize it. So, we bring NoStall!

What it does

NoStall lets users enter tasks as voice recording or just text and does the following:

1. Extracts a two minute version of the task - Easy version less inertia to start [Think: Open and read the title instead of read a research paper]
2. Extracts location and date-time based on task if no date/time is given
3. Prioritizes tasks based on deadlines dynamically
4. Location based task prioritization [Think: On the way to class, drop off the book you've been meaning to return]
5. Track your progress and stay motivated!

How we built it

We used a bunch of APIs and LLMs. Whisper is used for voice to text, LLAMA interact for extracting location and possible datetime [Both from Cloudfare], LLAMA versatile [from Groq] for prioritization. Our prioritization algorithm intelligently organizes tasks based on deadlines, urgency, and user preferences, ensuring optimal task management. Additionally, we’ve incorporated location-based task recommendations using GPS data, enabling the application to suggest relevant tasks from the list in real-time—similar to a game’s subquest that triggers when you reach a specific location. We used flask as the backend and html/css/js frontend. We deployed with render and used CloudFare DNS to route it to our .tech domain.

Challenges we ran into

Building this application wasn’t without its hurdles. One of the key obstacles was accurately breaking down large, often vague tasks into clear, actionable 2-minute subtasks while preserving the original context. Another major challenge was designing the prioritization algorithm, which needed to dynamically adjust based on deadlines and urgency. To refine this, we experimented with various models and engaged in prompt engineering. Additionally, we had to ensure that the system could reliably detect the user's location and provide tasks that were relevant to their current environment and goals.

Accomplishments that we're proud of

NoStall breaks down complex tasks into manageable 2-minute steps, making them less daunting and easier to handle. Its dynamic prioritization system ensures that users always focus on the right task, taking into account deadlines, urgency, and personal preferences. The app also offers location-based task suggestions, similar to game subquests, adding a personalized element to task management. With the voice-to-text feature, users can effortlessly input tasks, enhancing the app’s overall ease of

What we learned

We explored various techniques, ranging from graph algorithms for prioritizing tasks based on location to leveraging multiple GenAI models for smart, intelligent task management.

What's next for NoStall

The prioritization algorithm since it uses an LLM behind will eventually not scale after a while. We were working on a graph based prioritization with just LLM augmenting the features. NoStall right now has a lot of features that can be fully enhanced. We also plan to integrate with google calender.

Built With

• cloudfare
• css
• firestore
• flask
• geopy
• groq
• html
• javascript
• llama
• python
• render
• whisper