Inspiration
What it does# My Project: Building an Omi Plugin for Meaningful Conversation Analysis
What Inspired Me
The inspiration for this project came from observing the challenges faced by professionals like startup founders, investors, doctors, and students in managing their day-to-day conversations.
- Startup Founders struggle to keep track of the constant feedback they receive.
- Investors juggle multiple meetings and ideas, finding it difficult to recall critical points later.
- Doctors deal with complex patient discussions, making it crucial to extract actionable insights from their conversations.
- Students often need help organizing and summarizing lecture transcripts.
I wanted to create a tool that would help these individuals process their conversations efficiently and generate actionable outcomes, simplifying their decision-making process.
What I Learned
This project taught me a lot about user-centric design and leveraging advanced APIs to handle real-world challenges.
- Understanding User Needs: I learned to ask the right questions to identify what users truly need from the tool.
- AI-Powered Insights: Working with APIs like Gemini, I discovered how to process larger token lengths for better conversation analysis.
- Problem-Solving: Each user group (founders, investors, doctors, students) had unique challenges, which required tailoring the plugin's functionality to serve them effectively.
How I Built the Project
User Research:
- I began by identifying the target audience and designing an interactive flow to ask users about their profession and the kind of conversations they have.
- This flow guided the plugin to determine relevant key questions.
- I began by identifying the target audience and designing an interactive flow to ask users about their profession and the kind of conversations they have.
Core Functionality:
- The plugin leverages the Gemini API to analyze entire conversation transcripts using extended token limits.
- Based on the user's role, it generates actionable outcomes like summaries, categorized feedback, and task lists.
- The plugin leverages the Gemini API to analyze entire conversation transcripts using extended token limits.
Customization:
- Each profession gets a unique set of outcomes. For instance:
- Founders receive categorized feedback summaries.
- Investors get prioritized action points for ideas pitched.
- Doctors receive patient summary reports.
- Students receive structured lecture notes or study plans.
- Founders receive categorized feedback summaries.
- Each profession gets a unique set of outcomes. For instance:
Technology Stack:
- APIs: Gemini for natural language processing.
- Frameworks: Omi plugin environment for seamless integration.
- Frontend: Interactive, user-friendly interface to prompt users effectively.
- APIs: Gemini for natural language processing.
Challenges Faced
Long Token Limit Processing:
- Managing large transcripts efficiently required fine-tuning API requests and optimizing token usage.
- Managing large transcripts efficiently required fine-tuning API requests and optimizing token usage.
Personalization for User Roles:
- Ensuring actionable outcomes were relevant and useful to diverse professions was a complex task.
- Ensuring actionable outcomes were relevant and useful to diverse professions was a complex task.
Seamless Interaction:
- Designing a flow that felt intuitive and helpful without overwhelming the user was tricky.
- Designing a flow that felt intuitive and helpful without overwhelming the user was tricky.
API Rate Limits:
- Handling rate limits and ensuring fast response times in the Omi plugin environment required balancing performance with functionality.
- Handling rate limits and ensuring fast response times in the Omi plugin environment required balancing performance with functionality.
Outcome
The final product is a plugin that empowers users to turn conversations into actionable insights effortlessly. It not only saves time but also ensures professionals stay organized and informed, no matter how hectic their schedules are.
How We Built It
We started by identifying the pain points for professionals across different fields, such as startup founders, investors, doctors, and students. The development process included the following steps:
User Research and Flow Design:
- Designed a questionnaire to understand users' professions and the types of conversations they have.
- Mapped out workflows tailored to each profession's needs.
- Designed a questionnaire to understand users' professions and the types of conversations they have.
Technology Stack:
- Gemini API: Used for processing large conversation transcripts and extracting actionable insights.
- Omi Plugin Environment: Chose Omi for seamless integration into professional workflows.
- Frontend Development: Built an intuitive UI for user prompts and presenting outcomes.
- Gemini API: Used for processing large conversation transcripts and extracting actionable insights.
Functionality Development:
- Implemented personalized output logic for different roles.
- Designed the plugin to process transcripts efficiently using extended token limits.
- Focused on actionable outcomes like categorized summaries, decision points, and structured notes.
- Implemented personalized output logic for different roles.
Testing and Feedback:
- Ran iterative testing cycles to refine the plugin’s usability and accuracy.
- Incorporated feedback from real-world professionals to improve its practicality.
- Ran iterative testing cycles to refine the plugin’s usability and accuracy.
Challenges We Ran Into
Token Length Management:
- Balancing large conversation transcripts with API token limitations was a technical challenge.
- Balancing large conversation transcripts with API token limitations was a technical challenge.
Role-Specific Personalization:
- Developing custom outcomes for varied professions required deep understanding and dynamic logic.
- Developing custom outcomes for varied professions required deep understanding and dynamic logic.
API Rate Limits:
- Managing Gemini API’s rate limits while ensuring smooth functionality in the Omi environment was a constraint.
- Managing Gemini API’s rate limits while ensuring smooth functionality in the Omi environment was a constraint.
User Interaction Flow:
- Ensuring the question prompts were engaging yet concise without overwhelming users.
- Ensuring the question prompts were engaging yet concise without overwhelming users.
Accomplishments That We're Proud Of
Dynamic Role Personalization:
- Successfully created tailored outputs that cater to diverse professional needs.
- Successfully created tailored outputs that cater to diverse professional needs.
Seamless Integration:
- Built a plugin that integrates naturally into users' workflows, providing actionable insights with minimal effort.
- Built a plugin that integrates naturally into users' workflows, providing actionable insights with minimal effort.
Scalability:
- Designed the system to handle various user types and conversation complexities.
- Designed the system to handle various user types and conversation complexities.
Real-World Impact:
- Delivered a tool that can significantly reduce cognitive load and improve productivity for professionals.
- Delivered a tool that can significantly reduce cognitive load and improve productivity for professionals.
What We Learned
Importance of Personalization:
- Customizing outputs based on user roles enhances the tool's effectiveness and relevance.
- Customizing outputs based on user roles enhances the tool's effectiveness and relevance.
Optimizing API Usage:
- Learned strategies for working with API token limits and rate constraints to manage large data sets.
- Learned strategies for working with API token limits and rate constraints to manage large data sets.
User-Centric Design:
- Creating an intuitive and engaging user experience is as critical as building robust functionality.
- Creating an intuitive and engaging user experience is as critical as building robust functionality.
Iterative Development:
- Continuous feedback and testing are vital to refining the product and ensuring real-world applicability.
- Continuous feedback and testing are vital to refining the product and ensuring real-world applicability.
What's Next for Work Assist
Enhanced Analytics:
- Incorporate deeper sentiment analysis and trend identification in conversations.
- Incorporate deeper sentiment analysis and trend identification in conversations.
Multi-Platform Integration:
- Expand compatibility with other environments like Slack, Teams, and email platforms.
- Expand compatibility with other environments like Slack, Teams, and email platforms.
Advanced Customization:
- Allow users to customize the types of insights they want to see based on their workflow.
- Allow users to customize the types of insights they want to see based on their workflow.
AI Model Improvements:
- Explore fine-tuning the underlying AI models to better suit specific professional niches.
- Explore fine-tuning the underlying AI models to better suit specific professional niches.
Log in or sign up for Devpost to join the conversation.