AI Agent Automation Platform is an open-source, backend-first AI workflow automation system for building, running, and scheduling intelligent agent-driven pipelines — fully local, secure, and production-oriented.
This project is designed as a real execution engine, not a demo or prompt playground. It enables developers to orchestrate autonomous AI agents, automate backend workflows, run tasks on schedules, and inspect executions with full observability.
AI Agent Automation Platform is a self-hosted alternative to tools like n8n or Zapier, purpose-built for AI-driven automation. Everything runs locally or on your own infrastructure, ensuring data privacy, security, and full control.
Key ideas:
- Autonomous AI agents
- Deterministic workflow execution
- Secure local data processing
- Developer-friendly architecture
- Step-by-step deterministic execution
- Pluggable LLM adapter (OpenAI, Gemini, Groq, HuggingFace, etc.)
- Structured inputs & outputs per step
- Centralized executor with safety guards
- Step-level success and failure tracking
-
Visual workflow pipeline UI
-
Workflows defined as ordered steps
-
Supported step types:
- LLM reasoning
- HTTP requests
- Email automation
- File read / write / append
- Browser automation (sandboxed)
- Delay & control steps
- Cron-style scheduling
- Automatic task generation
- Supports recurring and long-running workflows
- Ideal for monitoring, reporting, and background automation
- Upload PDF documents
- Text extraction and semantic chunking
- Custom local vector store (no paid embedding APIs)
- Context-grounded answers only
- No hallucinations by design
- Task lifecycle tracking
- Step-level execution logs
- Timestamped results
- Real-time log viewer UI
- Persistent memory per agent
- Store facts, preferences, and prior context
- Enables adaptive and personalized automation
Frontend (Next.js + React)
↓
REST API (Express.js)
↓
Workflow Engine
├── Agent Runner
├── Step Executor
├── Tool Registry
├── Scheduler
└── Logger
↓
MongoDB (Tasks, Workflows, Logs, Agents)
This architecture enables:
- Clear execution boundaries
- Horizontal scalability
- Secure tool execution
- Full auditability
- Node.js
- Express.js
- MongoDB
- PM2
- Cron Scheduler
- Custom Agent Runner
- Next.js
- React
- Tailwind CSS
- DaisyUI
- Pluggable LLM adapters
- Custom execution pipeline
- Tool sandboxing
- Local vector store
- AI-powered workflow automation
- Scheduled reporting agents
- Backend task automation
- Intelligent document analysis
- Internal developer tools
- Secure AI experimentation platform
- Fully self-hosted (local or private cloud)
- No data leaves your system by default
- Secrets stored only in environment variables
- Sandboxed tool execution
- No vendor lock-in
git clone https://github.com/vmdeshpande/ai-agent-automation.git
cd ai-agent-automationcd backend
npm install
cp .env.example .env
npm run dev
npm run workerBackend runs at: http://localhost:5000
cd frontend
npm install
npm run devFrontend runs at: http://localhost:3000
backend/
├── agents/
├── controllers/
├── routes/
├── services/
├── models/
├── tools/
└── app.js
frontend/
├── app/
├── components/
├── dashboard/
└── styles/
- OCR support for scanned PDFs
- Public REST API for workflows
- Agent memory expansion
- Real-time updates (WebSockets)
- Plugin-based tool system
- Role-based access control
- This project follows semantic versioning.
- Current version: v0.1.0 (pre-1.0, APIs may change)
Contributions are welcome!
If you’re interested in AI agents, workflow automation, backend systems, or developer tooling, feel free to contribute.
See CONTRIBUTING.md for guidelines.
Licensed under the Apache License 2.0.
See the LICENSE file for details.
This is not a prompt playground.
This is a production-grade AI execution engine.