FocusLlama

Inspiration

have you ever tried to focus on fixing a bug but found yourself browsing for a yoga mat on Amazon? (You don’t even do yoga) have you ever googled a llama population in Tibet during daily meeting? we got you.

Meet FocusLlama, the ultimate solution for individuals battling distractions during online work sessions. Inspired by our personal journey with attention deficit and the alarming rise in diagnosed deficit disorders (which have increased by over 45% in the last 8 years), FocusLlama is here to make a difference.

What it does

In today’s fast-paced digital world, maintaining focus can be a real challenge. Many find themselves constantly drifting away from their intended tasks without even realizing it. That’s where FocusLlama steps in.

Our innovative browser extension actively monitors users’ focus, ensuring they stay on track with their initial session goals. We use Llamaindex + Groq + Llama 3 70B to detect categorize user content and detect distractions. If the user attempts to distract, FocusLlama provides gentle yet effective interventions-whether it’s a personalized pop-up message, the appearance of an adorable yet firm llama, or the closure of distracting tabs. In order to prevent overwhelming we suggest blurring off-topic photo and video on websites!

But that’s not all. With FocusLlama, users have access to comprehensive statistics tracking their total focused time,the frequency of distractions, and personalised optional time for work sessions based on the previous tasks. From daily insights to yearly trends, users can gain valuable insights into their productivity habits.

Don’t let distractions hinder your success. Join the FocusLlama revolution and take control of your focus today!

How we built it

Flask App with Llamaindex + Groq integrations for calling Llama3 70b with JSON schema formatting. We use it for classification. indexing and embedings. Custom scoring system to detect distraction.

Typescript + React Chrome extension to track all use events, IndexedDB for events storing, Supabase as a DB.

Swift widget to show notifications.

Challenges we ran into

JavaScript is a pain, especially on the chrome extension side, we have encountered several problems how we can collect the maximum number of events about user actions and how to process and store them. We managed to solve this with IndexedDB + Llamaindex + Grok, we categorize all user events before storing them in DB.

We also started using local Ollama to build embeddings to user events, which was new to us.

Accomplishments that we're proud of

We're excited to present the ADHD-focused browser extension, uniquely crafted to engage directly with the user. Unlike other extensions that simply block chosen sites, FocusLlama interacts directly with users to ensure their focus during work sessions.

(Chrome has a rather closed API and many things were a pain to work through)