Companio

Your AI companion for everyday conversations - who listen and don't judge!

Comment

Companio: Your AI Journaling Companion

Inspiration

In today's fast-paced world, many people struggle to maintain consistent journaling habits despite recognizing the mental health benefits. Traditional journaling requires dedicated time to sit down and write, which can be challenging to incorporate into busy schedules. We were inspired by the idea that voice is the most natural form of human expression, and that a phone call—something we do regularly—could be transformed into a journaling experience. By removing the friction of writing and making journaling as simple as a phone conversation, we aimed to make mental wellness practices more accessible to everyone.

What it does

Companio is an AI journaling companion that allows users to journal through natural phone conversations. Here's how it works:

  1. ElevenLabs agent initiates call to user via Twilio at set schedule and frequency as per user's requirements
  2. They're connected to an ElevenLabs AI voice agent that engages them in a natural conversation
  3. The AI asks thoughtful questions to help users reflect on their day, feelings, and experiences
  4. After the call, the conversation is extracted using ElevenLabs endpoint API and processed and summarized using Perplexity
  5. The journal entry and its summary are stored in Google Sheets for easy access and reflection
  6. Users can review their journal entries and track patterns in their thoughts and emotions over time

The system handles everything from the conversation to the summarization and storage, making journaling effortless and accessible.

How we built it

We built Companio using a multi-layered architecture that integrates several technologies:

  1. Voice Interface: We used Twilio to handle phone calls and ElevenLabs' voice AI technology to create natural, engaging conversations.

  2. Backend Processing:

    • A Node.js server handles webhooks from ElevenLabs and Twilio
    • Python scripts interacts with the ElevenLabs Get Conversation endpoint and process the conversation data
    • Perplexity API generates concise summaries of journal entries

  3. Data Storage: Google Sheets API stores the journal entries and summaries.

  4. Integration Layer: Custom webhooks connect all components, ensuring data flows smoothly from the phone call to storage.

The system is designed to be modular, with each component handling a specific part of the journaling workflow.

Challenges we ran into

Building Companio presented several challenges:

  1. API Integration Complexity: Integrating multiple APIs (ElevenLabs, Twilio, Perplexity, Google Sheets) required careful handling of authentication, rate limits, and data formats.

  2. Conversation Quality: Creating an AI voice agent that asks meaningful questions and responds empathetically required extensive prompt engineering and testing.

  3. Webhook Reliability: Ensuring reliable delivery and processing of webhooks between services was challenging, especially with varying response times.

  4. Summarization Accuracy: Generating concise yet meaningful summaries of open-ended journal conversations required fine-tuning the prompts and parameters.

  5. Error Handling: Building robust error handling across multiple services and languages (Node.js and Python) required comprehensive testing and fallback mechanisms.

Accomplishments that we're proud of

Despite the challenges, we achieved several significant accomplishments:

  1. Seamless User Experience: Creating a journaling experience that feels like talking to a friend rather than an AI.

  2. Cross-Platform Integration: Successfully integrating voice, telephony, AI, and data storage platforms into a cohesive system.

  3. Intelligent Summarization: Developing a summarization system that captures the essence of journal entries while highlighting key emotions and insights.

  4. Scalable Architecture: Building a system that can handle multiple users and scale as needed.

  5. Accessibility: Making journaling accessible to people who might not otherwise maintain a journal due to time constraints or the friction of writing.

What we learned

This project provided valuable learning experiences:

  1. Voice AI Capabilities: Gained deep insights into the capabilities and limitations of current voice AI technologies.

  2. Webhook Architecture: Learned best practices for designing reliable webhook-based systems that connect multiple services.

  3. Prompt Engineering: Developed skills in crafting effective prompts for both conversational AI and summarization tasks.

  4. Error Resilience: Learned to build systems that gracefully handle failures at various points in the processing pipeline.

  5. User Experience Design: Gained experience in designing voice-based user experiences that feel natural and engaging.

What's next for Companio

We have several exciting directions for Companio's future:

  1. Personalized Insights: Implementing advanced analytics to provide users with personalized insights about their emotional patterns and well-being.

  2. Multi-Modal Access: Expanding beyond phone calls to allow journaling through text messages, voice assistants, and a dedicated app.

  3. Custom Voice Agents: Allowing users to customize their AI journaling companion's personality, questioning style, and voice.

  4. Integration with Health Apps: Connecting with health and wellness apps to provide a more holistic view of mental and physical well-being.

  5. Guided Journaling Themes: Introducing specialized journaling sessions focused on gratitude, goal-setting, stress reduction, or other themes.

  6. Enhanced Privacy Features: Implementing additional privacy and security features to protect sensitive journal content.

By continuing to develop Companio, we aim to make mental wellness practices more accessible and effective for everyone, regardless of their schedule or journaling experience.

Built With

Share this project:

Updates