Callog

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

• 

Inspiration

We all have days where we just need someone to talk to — not a therapist, not a friend who might judge, just someone who listens. But journaling, the most recommended self-care habit, has an ironic problem: the people who need it most are the ones least likely to do it. Writing feels like a chore after a long day.

We asked ourselves: What if you could just talk about your day, and your diary wrote itself?

That's how Callog was born — an AI companion disguised as a fox who calls you, chats with you, remembers you, and turns your conversations into beautifully organized diary entries, emotional insights, and mental health reports. No writing required.

What it does

Talk, and your diary is already written.

Callog lets users have real-time voice calls or text chats with an AI fox companion. After each conversation, the app automatically generates:

• A diary entry summarizing the day's events
• Emotion analysis detecting how the user felt throughout the conversation
• A gratitude journal highlighting positive moments
• A depression score tracking mental health trends over time
• "How the fox sees you" — a unique AI perspective on the user's state

The fox remembers you. It recalls past conversations, asks follow-up questions about things you mentioned before, and grows to understand you over time. It's not just a chatbot — it's a companion that evolves with you.

Users also receive:

• Periodic letters from the fox with encouragement and reflection
• Emotion graphs visualizing mood patterns over weeks and months
• Feedback reports with strengths and areas for personal growth
• Category-based analysis breaking down recurring themes in their life

How we built it

Frontend: Built with Flutter for cross-platform iOS and Android support. The app features real-time voice recording with waveform visualization, live conversations, and a warm cream-and-coral UI designed to feel like a personal journal.

Backend: Serverless architecture powered by AWS Lambda (Python) with Amazon API Gateway. Over 20 Lambda functions handle everything from diary CRUD operations to AI-powered analysis and subscription management.

AI & Conversation:

• Gemini API powers the core conversational AI, enabling natural, context-aware dialogue
• Voice conversations use real-time audio streaming with configurable voice personalities (sage, ash, ballad, coral, verse) and adjustable tone/speed
• Conversation analysis pipeline processes each session through multiple AI passes: summarization, emotion extraction, gratitude identification, and psychological assessment

Database: Amazon DynamoDB stores diary entries, conversation messages, AI analysis results, and user profiles with efficient partition key design (userId#timestamp).

Authentication: Firebase Authentication with Email/Password, Google Sign-In, and Sign in with Apple. Firebase ID tokens secure all API calls.

Monetization: In-app subscriptions via App Store and Google Play, with server-side receipt validation through dedicated Lambda functions.

Internationalization: Supports 16 languages including English, Korean, Japanese, Chinese, Spanish, French, German, and more — making mental wellness accessible globally.

Challenges we ran into

Making AI feel like a friend, not a bot. The hardest challenge wasn't technical — it was emotional. We spent significant time tuning conversation prompts and voice parameters so the fox feels warm, genuine, and personal. A mental health companion that feels robotic does more harm than good.

Real-time voice conversation reliability. Handling connections for live voice diary sessions across varying network conditions required careful error handling, reconnection logic, and audio stream management.

Memory across sessions. Making the fox "remember" users meant designing a system that retrieves and contextualizes past conversations without overwhelming the AI's context window. We built a retrieval pipeline that surfaces relevant past entries during new conversations.

Cross-platform subscription management. Handling iOS App Store and Google Play subscriptions with different product IDs, receipt formats, and edge cases (expired subs, account conflicts, restoration) in a unified system was a significant engineering challenge.

Accomplishments that we're proud of

• Users consistently report that talking to the fox feels more natural than writing a diary — our core hypothesis validated
• The automatic diary generation accuracy is high enough that most users don't edit their entries
• Successfully launched on both App Store and Google Play with real paying subscribers
• Supporting 16 languages from day one, making mental health tools accessible to a global audience

What we learned

• The best UX is no UX. The less the user has to "do," the more they engage. Talking is effortless; the AI handles the rest.
• Emotional AI requires emotional design. Every micro-interaction — the fox's tone, the diary's layout, the notification timing — affects whether users trust the companion.
• Serverless scales beautifully for diary apps. Lambda's pay-per-use model perfectly matches the bursty usage pattern of a daily diary app.

What's next for Callog

• Deeper memory and personality modeling — the fox will develop a richer understanding of each user's personality, communication style, and life context
• Group journaling — shared diary spaces for couples, families, or friend groups
• Professional integration — optional sharing of anonymized insights with therapists or counselors
• Proactive check-ins — the fox reaches out when it senses the user might need support based on mood trends

Built With

• amazon-api-gateway
• amazon-dynamodb
• apple
• aws-lambda
• dart
• firebase-analytics
• firebase-authentication
• flutter
• gemini-api
• google-play-billing
• in-app-purchase
• python
• storekit
• websocket