Core Mechanism¶

Bridgic lets you build agentic systems by breaking down your workflows into modular building blocks called worker. Each worker represents a specific task or behavior, making it easy to organize complex processes.

Bridgic introduces clear abstractions for structuring flows, passing data between execution units, handling concurrency, and enabling dynamic control logic (such as conditional branching and routing). This design allows users to build systerm that scale efficiently from simple workflows to sophisticated agentic systems.

Key features include:

1. Concurrency Mode: Organize your concurrent execution units systematically and conveniently.
2. Parameter Resolving: Explore three ways for passing data and two ways for dispatching data between execution units, including Arguments Mapping, Arguments Injection, and Inputs Propagation about parameter binding; including Input Dispatching and Result Dispatching.
3. Dynamic Routing: Enable conditional branching and decision-making through an easy-to-use ferry_to() API.
4. Dynamic Topology: Change the underlying graph topology at runtime to support highly autonomous AI applications..
5. Human-in-the-loop: Enable human interaction or external input during workflow / agent execution.
6. Worker Callback: Bring non-intrusive and multi-scope callback mechanisms that allows you to observe, customize, or extend execution at different levels with minimal impact on core logic.
7. Modularity: Reuse and compose Automata by embedding one inside another for scalable agentic systems.
8. Model Integration: Incorporate model to building a program with more autonomous capabilities.

This architectural foundation makes Bridgic a powerful platform for building agentic systems that are robust, adaptive, and easy to reason about, enabling you to bridge the precision of logic with the creativity potential of AI.