Core Concepts of Operaide
Operaide is a platform for building, deploying, and running AI-powered applications. It is designed to be simple and focused. A small set of concepts is all you need — whether you are designing processes, building apps, or managing deployments.
Key Concepts
| Concept | Description | |
|---|---|---|
| 🤖 | AI Integration | Native support for multiple AI providers (OpenAI, Azure, Anthropic, etc.) |
| 🧩 | Composable Architecture | Build complex workflows from simple, reusable components |
| 💻 | Code-Centric Development | Write workflows in TypeScript with full type safety |
| 🛠️ | Browser-based IDE | For building, testing, and deploying apps without local setup and configuration |
| ✨ | Vibe Coding/Engineering | AI-driven development with complete Operaide coding context |
| 🔌 | REST API | Deploy workflows as scalable REST endpoints |
| 🏪 | App Store | Discover and deploy ready-made Apps, use them as templates, or contribute your own |
| 🗄️ | Database | Built-in database and vector storage for persistent data and RAG applications |
| 📄 | Document Intelligence | Process, analyze, and extract structured information from documents |
| 💬 | Chat | Build custom chat interfaces powered by your workflows |
| 🖥️ | User Interface | Create your own business UI within your App |
| 🔐 | Roles & Access Control | Fine-grained permissions per user, organisation, and App Instance |
| 📊 | Monitoring & Reporting | Real-time usage reports per instance |
📦 The App — Where Your Business Logic Lives
Every workflow in Operaide lives inside an App. It is the central unit of the platform — a self-contained package that brings together your business logic, an optional custom interface, and everything else needed to run.
⚡ Business Logic: Reaktors and Aktors
The logic inside an App is organised into Reaktors. Each Reaktor represents one workflow — a document processing pipeline, a customer support agent, a data extraction job. Reaktors expose themselves as REST endpoints, ready to be called from anywhere.
Inside each Reaktor, Aktors do the actual work. They are small, composable functions — one calls an AI model, another transforms data, a third sends an email. Aktors chain together to form the full execution flow, and every step is traceable.
Operaide uses the German spellings Reaktor and Aktor rather than the English "Reactor" and "Actor". The platform was built by a German-speaking team and the names have stayed — you will see them consistently throughout the codebase and all documentation.
🚀 One App, Many Configurations
The same App can run as multiple independent App Instances — each with its own settings. Point one instance at GPT-4o and another at Claude. Run the same invoice-processing App for ten different clients, each fully isolated. Change the system prompt per customer without touching a single line of code.
🎛️ Optional: UI and Database per Instance
Apps are not limited to backend logic:
- 🖥️ App UI — embed a custom front-end directly into your App. Users get a tailored interface; you maintain a single codebase.
- 🗄️ Agent DB — each App Instance can have its own database and vector store, purpose-built for AI workflows. Data stays isolated between instances — ideal for multi-tenant scenarios.
- ✅ Tests — Apps can include automated quality checks that verify the behaviour of your Reaktors. Run them on every deployment to catch regressions before they reach production.
The diagram below shows all the pieces.
