Inspiration
The inspiration for Omi came from the creator's personal struggle with the traditional process of online research and content management. This process often involved opening numerous tabs, manually saving relevant information, and struggling to recall where certain pieces of information were found. The realization that these pain points are common among knowledge workers led to the idea of Omi, a smart assistant that could simplify and streamline the entire process.
What it does
Omi is a Chrome extension that functions as a personal research assistant. It remembers every web page a user visits, saving the URL and content. It not only remembers information, but also allows the user to interact with it. The user can ask Omi questions about the materials they've read, and it will provide answers along with references. Omi can also assist in content generation, sourcing from the websites the user has visited. All data is stored locally, ensuring privacy and security.
How we built it
Omi was built using OpenAI's GPT-3.5. Once the extension is activated, it tracks all the webpages visited by the user for 90 minutes, storing the content and details locally. The user can then interact with the extension's chat feature to ask questions regarding the materials they have read. Omi retrieves the relevant information along with its source.
Challenges we ran into
The initial idea was to create a productivity software that would help people create PowerPoint presentations, but this idea was deemed too narrow. The challenge was to expand the scope of the application to encompass general research and content generation, not just presentation creation.
Accomplishments that we're proud of
The development of a lightweight, secure, and user-friendly Chrome extension that drastically improves the online research process is a significant achievement. Omi's ability to recall and reference information eliminates the common problem of forgetting where specific information was found.
What we learned
The development process provided valuable insights into the limitations of traditional research methods and the potential of AI-powered solutions. It also highlighted the importance of understanding the needs of the target audience and ensuring that the product can be adapted to suit broader applications.
What's next for Omi
The future plans for Omi involve transforming it into a cross-platform and cross-device product that can handle various forms of data generated throughout the day, including emails, calendars, text messages, voice memos, and more. This will be done using in-context learning combined with the user's personal data. Omi's developers also plan to start building their own language learning model to improve results and reduce costs. They aim to leverage Omi's existing user base to drive growth for the new, enhanced platform.
Built With
- prompt
- typescript
- vector-db
Log in or sign up for Devpost to join the conversation.