AI-powered personal ops engine for builders who move fast.

Problem

Modern builders lose time across disconnected chat tools, scripts, lead workflows, and manual follow-ups. Context gets fragmented, execution slows down, and high-value actions stay stuck in draft mode.

Solution

Rock Personal Assistant brings chat, memory, retrieval, lead generation, and controlled outbound workflows into one modular system. You get a simple interface with real-time execution, while keeping human confirmation for high-risk actions like message sending.

Features

• AI-powered multi-model chat with persistent session memory
• Real-time Telegram + web interfaces sharing one brain
• Modular orchestration APIs for plan -> execute -> confirm flows
• Lightweight RAG with local document ingestion and retrieval
• Controlled outbound pipeline with explicit send confirmation
• SQLite-backed state for runs, tasks, actions, and contacts
• Production-friendly structure with clear extension points

Architecture

Rock is designed as modular layers so each part can evolve independently.

flowchart LR
    A["Web UI / Telegram"] --> B["FastAPI Gateway"]
    B --> C["AI Runtime"]
    C --> D["Memory + RAG (SQLite)"]
    B --> E["Rock Orchestrator"]
    E --> F["Lead Generation Operator"]
    E --> G["OpenClaw WhatsApp Agent"]
    E --> H["Confirmation Guardrail"]
    H --> G

Loading

Use Cases

• Solo founder running lead research + draft outreach from one interface
• Developer assistant for quick reasoning, context memory, and document Q&A
• Team operator mode where AI drafts actions and humans approve final sends
• Personal productivity stack combining chat, notes, summaries, and automation

Tech Stack

• Backend: Python, FastAPI, Uvicorn
• AI/LLM: NVIDIA API models, prompt routing
• Data: SQLite, local document chunking
• Integrations: Telegram Bot API, OpenClaw bridge
• Frontend: Static HTML/CSS/JS

Installation (Linux)

git clone https://github.com/SaieshwarTech/rock-personal-assistant.git
cd rock-personal-assistant

python3 -m venv .venv
source .venv/bin/activate

pip install -r requirements.txt
cp .env.example .env

Set values in .env:

NVIDIA_API_KEY=your_nvapi_key_here
TELEGRAM_BOT_TOKEN=your_telegram_bot_token_here
TELEGRAM_BOT_ENABLED=true

Run:

uvicorn app:app --reload --port 8000

Open http://127.0.0.1:8000.

CLI/API Usage

This project is API-first for orchestration workflows.

# Execute an orchestrated task
curl -X POST http://127.0.0.1:8000/rock/execute \
  -H "Content-Type: application/json" \
  -d '{"session_id":"rock-main","user_text":"Find 10 gym owners and message them"}'

# Approve or reject outbound send
curl -X POST http://127.0.0.1:8000/rock/confirm-send \
  -H "Content-Type: application/json" \
  -d '{"run_id":"<run-id>","approved":true}'

# Fetch run timeline
curl http://127.0.0.1:8000/rock/runs/<run-id>

Telegram commands:

• /start, /help, /models, /model <model_name>
• /temp <0.0-1.5>, /memory on|off, /rag on|off
• /system <text>, /system_clear, /session, /newsession

Demo

$ curl -X POST /rock/execute ...
intent: lead-outreach
plan: collect leads -> draft messages -> await approval
execution: 10 leads found, 10 drafts prepared
result: pending confirmation (safe mode)

Dashboard preview:

• Web app at http://127.0.0.1:8000
• Telegram live assistant for on-the-go operations

Roadmap

• Multi-user workspaces with role-aware permissions
• Pluggable tool registry for custom operators
• Streamed live execution events over websockets
• Advanced analytics dashboard for run quality and outcomes
• Cloud deployment profile for scalable startup teams

Contributing

Contributions are welcome and encouraged.

1. Fork the repo
2. Create a branch: git checkout -b feat/your-feature
3. Commit changes: git commit -m "feat: your feature"
4. Push branch: git push origin feat/your-feature
5. Open a Pull Request

Please read CONTRIBUTING.md, CODE_OF_CONDUCT.md, and SECURITY.md before contributing.

License

Released under the MIT License. See LICENSE.