Comrade AI

The friend who listens, understands, and remembers.

Comrade AI is an AI journaling companion that learns from your writing and provides personalized emotional support through intelligent journaling, voice conversations, text chat, and emotional insights.

Features

Intelligent Journaling (/write)

Rich markdown editor (Milkdown) with auto-save, mood tagging, and a searchable archive. Journal entries are semantically indexed via SuperMemory so the AI understands your history and emotional patterns over time.

Voice Conversations (/talk)

Real-time voice calls powered by LiveKit. Choose from multiple AI speaker personas (Abhimanyu, Arjuna, Mira, Gargi, Karna, Kavya). A separate Python agent (voice-agent/) runs the speech pipeline entirely on Groq's free tier — no paid speech provider needed. Includes a per-user quota system (5 minutes free).

AskComrade Chat (/chat)

Text-based conversations backed by memory-aware context (Groq openai/gpt-oss-120b). The AI recalls relevant journal entries and your profile to provide personalized responses. Conversations are persisted with auto-generated titles.

Mind Graph (/mind)

Visual knowledge graph of your memories and emotional patterns using SuperMemory's memory graph component.

Tech Stack

Layer	Technology
Framework	Next.js 16 (App Router), React 19, TypeScript
API	tRPC 11 with React Query
Database	PostgreSQL (Neon serverless) + Drizzle ORM
Auth	Clerk (custom flows + bot sign-up protection)
LLM	Groq (openai/gpt-oss-120b)
Memory	SuperMemory (semantic storage & retrieval)
Voice	LiveKit Cloud + voice-agent/ (Groq STT whisper-large-v3-turbo, LLM gpt-oss-120b, TTS Orpheus)
Editor	Milkdown (markdown)
Payments	Dodo Payments
Styling	Tailwind CSS 4, shadcn/ui

Prerequisites

• Bun runtime
• Python 3.10+ (only for the voice agent)
• PostgreSQL database (recommended: Neon)
• Account keys for Clerk, Groq, SuperMemory, and LiveKit

Getting Started

1. Clone the repo and install dependencies:

git clone <repo-url>
cd comrade-ai
bun install

2. Copy the example env file and fill in your keys:

cp .env.example .env

Required variables:

Variable	Description
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY	Clerk frontend key
NEXT_PUBLIC_CLERK_SIGN_IN_URL	/sign-in
NEXT_PUBLIC_CLERK_SIGN_UP_URL	/sign-up
NEXT_PUBLIC_CLERK_AFTER_SIGN_IN_URL	/chat
NEXT_PUBLIC_CLERK_AFTER_SIGN_UP_URL	/onboarding
CLERK_SECRET_KEY	Clerk backend key
CLERK_WEBHOOK_SECRET	Clerk webhook signing secret (only needed if you deliver webhooks)
SUPERMEMORY_API_KEY	SuperMemory API key
DATABASE_URL	PostgreSQL connection string
GROQ_API_KEY	Groq API key (chat + voice)
CRON_SECRET	Secret for cron job endpoints
LIVEKIT_URL	LiveKit Cloud WSS URL
LIVEKIT_API_KEY	LiveKit API key
LIVEKIT_API_SECRET	LiveKit API secret
FEEDBACK_WEBHOOK_URL	Discord feedback webhook URL
DODO_PAYMENTS_API_KEY	DODO Payments API key
DODO_PAYMENTS_WEBHOOK_KEY	DODO Payments webhook key
DODO_PAYMENTS_RETURN_URL	http://localhost:3000
DODO_PAYMENTS_ENVIRONMENT	test_mode or development

3. Push the database schema:

bun run db:push

4. Start the dev server:

bun dev

The app will be available at http://localhost:3000.

Voice Agent

The /talk feature requires a separate Python worker that joins LiveKit rooms and runs the speech pipeline. See voice-agent/README.md for full setup and the character → voice mapping.

Quick start:

bun run voice:setup            # create venv + install deps
# fill voice-agent/.env (LIVEKIT_* — must match root .env — and GROQ_API_KEY)
bun run voice:dev              # run the agent in hot-reload dev mode

Run everything together: bun run dev:all (Next.js + voice agent via concurrently).

One-time setup: Groq's Orpheus TTS requires accepting the model terms at https://console.groq.com/playground?model=canopylabs%2Forpheus-v1-english while logged in, otherwise TTS fails with model_terms_required.

Scripts

bun dev                    # Start dev server (Next.js + Turbo)
bun run build              # Production build
bun run start              # Serve production build
bun run typecheck          # tsc --noEmit
bun run test               # Vitest
bun run db:push            # Push schema to database
bun run db:studio          # Drizzle Studio
bun run db:generate        # Generate migration
bun run db:migrate         # Run migrations
bun run voice:setup        # Set up the voice agent venv
bun run voice:dev          # Run the voice agent (dev mode)
bun run voice:console      # Run the voice agent in console mode
bun run dev:all            # Run Next.js + voice agent together

License

Private.