A visual flow represents an application as connected, Lego-like components. Each component performs a task, while input and output ports show how data or control moves through the workflow. section 2.5
Visual flows shift attention from syntax and framework details to design: select components, configure parameters, connect stages, and iterate quickly. The same architecture can later be translated into code. section 1.1 section 2.1
A typical GenAI flow can contain:
- An input component that receives the user message. section 2.5
- An LLM component that generates a response, guided by a system message. section 2.5
- An output component that presents the result. section 2.4
Pirate chatbot
Connect Chat Input to an OpenAI model and then Chat Output; add a system message requiring Pirate English. The resulting Langflow design corresponds directly to a LangChain chain written in Python. section 2.5 section 2.4
Figure 2.18 Meet Langflow, the wizard of GenAI apps.
Use the figure to orient the visual canvas as a workspace for assembling GenAI application components. section 2.4
