Introduction

AI Workflows in AI Squared help teams orchestrate dynamic, context-aware reasoning pipelines using a visual interface. These workflows combine chat input, data retrieval, LLM reasoning, and output visualization, helping teams to build everything from intelligent chatbots to fully embeddable analytics flows. Workflows are made up of modular components connected in a canvas-based builder, and can be triggered via chatbots.

What AI Workflows Enable

– Build retrieval-augmented generation (RAG) flows using vector search – Chain together chat input, prompts, LLMs, databases, and output UIs – Visualize responses as tables, charts, or rich summaries – Run workflows behind chatbots or assistants with no code – Reuse connectors from Data Movement or AI Activation

Key Concepts

ConceptDescription
Chat InputAccepts user questions to initiate the workflow
PromptTemplates used to structure input for the LLM
LLMConnects to OpenAI, Anthropic, etc., to generate completions
Database / Vector SearchQueries your structured or embedded data
Chat OutputDisplays the final result in the interface
Workflow CanvasVisual UI to arrange and connect components