Workspace Default Flow - GoInsight.AI

Overview

This template solves the headache of manually managing AI conversation modes by automating the decision to invoke either a search-enabled agent or a baseline dialogue agent. It streamlines workflow logic to dynamically route user queries based on front-end inputs, saving time and reducing errors while enabling advanced knowledge integration when needed.

The Impact

Slash manual routing. Automatically selects AI mode based on user preferences and context.
Boost response relevance. Enables dynamic web search integration for up-to-date answers.
Cut development overhead. Agent nodes handle tool invocations without coding complex logic.
Scale with ease. Supports future extensions like private knowledge bases and multi-tool workflows.

Who This Is For

Product Managers who require flexible AI workflows for diverse user queries.
AI Developers building conversational agents needing seamless tool integration.
Support Teams aiming to automate knowledge retrieval and FAQ handling.
Business Analysts who want to embed company context and private data dynamically.

How It Works

Start Node Captures Inputs
Receives user query plus front-end variables: Search (boolean) and Model (string) to guide the workflow.

Conditional Branching Logic
Checks Search variable: if true, routes to Search-Capable Agent; if false, routes to Baseline Agent.

Search-Capable Agent Execution
Uses ReAct framework to autonomously decide when to invoke search tools, integrate results, and refine responses.

Baseline Agent Execution
Handles direct Q&A with no external search, optimizing for quick, lightweight interactions.

Output Delivery
Each agent sends its response to the corresponding reply node, delivering the final answer seamlessly.

What You'll Need

Before using this template, make sure you have:

Access to GoInsight Workspace with the ability to edit and publish workflows.
Frontend interface with toggles for "Allow Web Search (Search)" and "AI Model Selector (Model)" to pass variables.
Configured credentials and permissions for any integrated search tools (e.g., Google Search API) if search is enabled.
Familiarity with workflow publishing to save and activate changes after editing.

How to Use

Step 1. Configure Frontend Selectors

Set up interface controls to allow users to enable/disable web search and select AI models, passing these as variables into the workflow.

Step 2. Edit Workflow Logic (Optional)

Adjust the conditional branching or agent parameters to fit specific business needs, like adding company context or private knowledge integration.

Step 3. Save and Publish

After any changes, save and publish the workflow to activate updates; otherwise, changes remain in draft mode.

Step 4. Test Conversation Starts

Initiate dialogues with varying Search toggle settings to verify the correct agent is invoked and responses are accurate.

Step 5. Monitor and Refine

Review interaction logs and user feedback to fine-tune prompt templates and agent configurations for optimal performance.

FAQs

How does the workflow decide which AI agent to use?

It uses the "Search" boolean variable from the frontend; if true, it invokes the Search-Capable Agent, otherwise the Baseline Agent.

What advantages do Agent Nodes provide over basic LLM nodes?

Agent Nodes encapsulate tool invocation logic, autonomously handling search calls, result integration, and response refinement without manual workflow coding.

Can this workflow be customized to include private company knowledge?

Yes, you can extend the workflow logic and agent capabilities to incorporate Retrieval-Augmented Generation (RAG) with your private knowledge base.

What happens if I forget to publish after editing the workflow?

Changes will remain saved as a draft and won’t affect active conversations until you explicitly save and publish the workflow.

Updated on: Aug 21, 2026

Was This Page Helpful?