HackStack

Project Info

Trayc

Devpost

This project did not submit a demo video on Devpost.

Inspiration

In our fast-paced lives, we have countless conversations and experiences that fade from memory. We were inspired to create a tool that helps anyone capture and recall the meaningful moments they might otherwise forget, from a busy professional needing to remember key discussion points to a grandparent wanting to cherish family moments. Trayc acts as a seamless, AI-powered extension of your own memory.

What it does

Trayc is an AI-powered memory companion that automatically documents your life's conversations and locations, creating a searchable map of your experiences. Here’s how it works It Listens & Locates: Trayc passively hears your conversations and notes your location throughout the day. It Connects the Dots: The AI seamlessly links every conversation to its specific location on an interactive map. It Talks Back: You can chat or speak directly with Trayc to ask questions about your past. Just ask, "What was I talking about when I was at the park yesterday?" It Replays Your Moments: Simply click on any point in the map to instantly see a summary of what you were discussing the last time you were there.

How we built it

We used Groq for the speed, Vapi for the intelligent voice agents, Python for the backend, Typescript and React for the frontend, and ChromaDB for the vector database.

Challenges we ran into

The location data wasn't being properly captured and passed through the entire pipeline. The problem was processing Vapi transcript events in real-time, so we used event listeners collecting transcript segments during calls, processing on call end with proper speaker diarization. Designing two rag systems: One for reading on screen for text-based input and one for hearing over the phone for audio input.

Accomplishments we're proud of

We were able to store location data corresponding to conversation data in a vector database We achieved sub-second retrieval using Groq. We're proud of building a system with privacy and user control at its core, from the architecture to the UI. We used ChromaDB, which has all safety protocols. Finetuned and engineered the AI to distinguish between a casual chat and an important work discussion

What we learned

We gained technical experience in real-time audio processing and vector databases. We also learned that the most impactful AI applications are those that intuitively augment human capabilities. We also learned that building trust is as important as building features when creating a product that handles sensitive personal data.

What's next

Trayc has an exciting journey ahead of it! This includes features such as the following: With Augmented Reality hardware, Trayc will be able to have a deep understanding of your brain and personality, and identify locations and/or situations where you tend to forget things. Trayc will provide a fun and interactive way to remember everything you need to know at any time. With Augmented Reality hardware, Trayc will be able to have a deep understanding of your brain and personality, and identify locations and/or situations where you tend to forget things. Trayc will provide a fun and interactive way to remember everything you need to know at any time. The ultimate goal for Trayc is to learn about you and your personality over time, and identify situations and/or locations where you forget things, and intelligently guide you in your life. It will provide real-time information, assistance, help recall memories and conversations, and also provide feedback for users. The ultimate goal for Trayc is to learn about you and your personality over time, and identify situations and/or locations where you forget things, and intelligently guide you in your life. It will provide real-time information, assistance, help recall memories and conversations, and also provide feedback for users.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • FirebaseIn code
  • HTMLIn code
  • JavaIn code
  • JavaScriptIn code
  • ReactIn code
  • Tailwind CSSIn code
  • TypeScriptIn code
  • PythonClaimed

8 of 9 appear in the indexed code. 1 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

355 KB

Source files

98

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars