RedoraAI is an open-source, AI-powered lead generation platform for Reddit.
It automates lead discovery, monitoring, and engagement — helping you find and connect with high-intent users across Reddit while staying fully compliant with community rules.

🚀 Automate your Reddit lead generation with AI agents that discover relevant subreddits, monitor discussions, and craft authentic, rule-safe engagement messages.

You can explore and use RedoraAI online at https://redoraai.com

• Features
• Architecture Overview
• Tech Stack
• Getting Started
  • Configuration
  • Backend Setup
  • Frontend Setup
  • Running the Project
• Integrations
• Deployment
• Account Safety Strategies
• Gotchas & TODOs
• Contributing
• License

• Extract relevant posts for given keywords, subreddits, and product details
• AI-assisted keyword and subreddit suggestions
• Intelligent post scoring using LLMs

• AI-generated DMs and comments tailored to community guidelines
• Scheduled auto-replies on relevant posts
• Auto DMs with configurable timing and frequency

• Email notifications via Resend
• Slack alerts for daily or weekly summaries
• Email Alerts to users when a connected account is revoked or banned
• Onboarding emails
• Subscription expiry and renewal alerts

• Connect multiple Reddit accounts
• Auto-rotation for comments and DMs
• Rotation strategies:
  • Random — pick a random account
  • Specific — use a chosen account
  • Most Qualified — based on karma, age, etc.

• Daily and weekly engagement summaries

• Simple in-app subscription logic with plan limits

• Manage all AI-generated comments and DMs

• Generate and schedule Reddit posts
• Posts follow subreddit rules and guidelines

This repository contains all components required to run RedoraAI.

.
├── backend/             # Go backend services
│   ├── portal-api/      # Public API layer for frontend (gRPC + Connect)
│   └── spooler/         # Core tracking engine for subreddits and posts
├── frontend/            # Frontend mono-repo
│   ├── portal/          # Web app (Next.js + PNPM)
│   └── packages/        # Shared UI, config, and protobuf packages
└── devel/               # Local development setup scripts

• Portal — API layer for frontend over gRPC/Connect-Web
• Spooler — Tracks relevant posts based on subreddit/keyword pairs every 24h
  • Configurable fanout limits, polling intervals, and daily quotas
  • Limits no of posts to track
  • Limits Max no of relevant posts per day as per the subscribed plan
  • Built-in retry logic for failed posts
  • Send scheduled posts, DM and Comments

• Powered by LiteLLM (deployed separately)
• Supports OpenAI and Gemini APIs interchangeably
• Handles scoring, comment/DM generation, and keyword/subreddit suggestions

Backend

• Go 1.23+
• PostgreSQL
• Redis
• Docker
• LiteLLM
• Playwright

Frontend

• Node.js 20+
• PNPM
• Next.js / React
• Tailwind CSS / Material UI

Auth & APIs

• Auth0 (passwordless login)
• Reddit OAuth APIs
• Resend (emails)
• Browserless / Steel.dev (CDP automation)
• DODO Payments (subscriptions)

Ensure you have installed:

• Docker
• Go 1.23+
• Node.js 20+
• PNPM
• direnv for environment variables

Start local PostgreSQL and Redis:

./devel/up.sh

Copy the environment file and configure:

cp .envrc.example .envrc
direnv allow

Replace placeholders (<value>) with your actual secrets and keys.

Run tests and start the backend:

cd backend
go test ./...
go build -o redora && ./redora start

Initialize the database:

./backend/script/migrate.sh up

Create a new migration:

./backend/script/migrate.sh new <migration_name>

Install dependencies:

cd frontend
pnpm install

Start the development server:

pnpm dev:portal

Visit: http://localhost:3000

You’ll need three components running:

1. Docker — for Postgres, Redis, Pub/Sub emulator

   ./devel/up.sh

2. Backend — use reflex for live reload

   reflex -c .reflex

3. Frontend — Next.js app

   cd frontend && pnpm dev:portal

Visit:

• http://localhost:8081 → pgweb (Postgres UI)
• http://localhost:3000 → Redora Portal

Integrations store external service credentials and configuration.

Manually insert an integration using tools. Example (CLI):

doota tools integrations slack_webhook create <org-id> '{"channel":"redora-alerts","webhook":"<slack-url>"}'

There is no separate admin interface. Users assigned the role PLATFORM_ADMIN can view all organizations and have access to all accounts across the platform.

• Hosted on Railway.app
• GCP used for LLM and Playwright storage
• Secrets managed via GCP KMS
• PostgreSQL and Redis hosted on Railway

To minimize Reddit account bans:

1. Age-based limits — automate only with accounts > 2 weeks old
2. Gradual scaling — increase activity slowly
3. DM-first approach — DMs are safer than comments
4. Rule adherence — generate context-aware replies
5. Consistent engagement — reply and post regularly
6. Occasional posting — maintain activity score

• We currently use OpenAI for keyword/subreddit suggestions during onboarding and LiteLLM for other AI related tasks. Ideally, we should use a single AI provider for all tasks.
• When scoring a post, if the score is >90, we double check it with an advance model. This could be improved by selecting a better default model.
• Comment and DM generation should be moved into separate LLM calls. Right now, scoring, comment and DM generation are all done in a single LLM call.
• We should add the ability to regenerate comments or DMs.
• To avoid getting banned, we only use Reddit accounts that are > 2 weeks old for AI generation. This is a temporary solution and we should come up with a better way to handle account warmup.

We welcome contributions!

1. Fork the repo
2. Create a new branch:

   git checkout -b feature/your-feature

3. Commit your changes
4. Open a Pull Request

Please follow Go and JS linting rules before submitting.

Released under the MIT License.
© 2025 DoneByAI — building AI tools that work for you.

• DoneByAI Team
• Shashank Agarwal