Shared State in Flowise is Literally Game Changing
AI Summary
This video provides a comprehensive overview of how to use shared state in Flowwise agent flows. It explains the concept of shared state as a powerful feature to save tokens, improve performance, and get better results from LLMs, contrasting Flowwise’s approach with traditional deterministic workflows. The presenter illustrates the use of conversation history as a form of shared state, noting its drawbacks such as growing token usage and declining model quality with longer contexts. They then introduce flow state as a solution — a system of unique state variables or containers accessible and updatable by different agents in a workflow. The example given is a blog creation flow with nodes generating a blog outline, post, and X/Twitter post, where each node accesses only the specific piece of state needed instead of the full conversation history. The video then shows live demonstrations of setting up and using flow state variables within Flowwise, including conditional logic based on state and using both LLM nodes and custom function nodes to update state variables. The benefit of flow state is emphasized in reducing token usage and improving response quality by passing only necessary information between nodes. The video concludes by encouraging users to like and subscribe for more Flowwise content and to ask questions about state in the comments.