Flow Editor

This guide walks you through the flow editor interface, explains the core principles of building flow, and shows you how to assemble the simplest chatbot possible. By the end of this guide, you’ll have a working chatflow and the confidence to expand it with more tools and data sources.

What are Flows

Before we learn how to build flows, we need to understand what flows are. You can think of flows as the brains behind your chatbots and AI tools. They are the strings connecting AI with various tools and capabilities in a logical sequence. Why Flows? Because the information flows through all the components, from the initial input to the final AI answer or or agent action.

What is The Flow Editor

The Flow Editor is a simple, visual workspace for building AI-powered workflows. Instead of writing code, you arrange building blocks on a canvas and link them so the information is processed and moves from input to outcome.

Each component performs a specific task—whether that’s taking user input, processing information, or generating AI responses. By connecting these blocks in logical sequences, you create powerful workflows without needing any programming knowledge.

How to Access The Flow Editor

How you access the builder differs slightly based on whether you’re building from scratch, editing an existing flow or creating a flow from the library.

• To start building from scratch, go to My Flows -> +Add New Flow -> + Create blank Flow
• To edit from library, pick the flow you want and click + Add to my flows
• To edit an existing flow, go to My Flows, hover over the desired flow and click the pencil icon.

Flow Editor Overview

Let’s try building from scratch. Open a blank flow. On entering the editor, you’ll see a simple canvas with Chat Input and Chat Output components already there to help you get started:

Top navigation bar

The top navigation bar allows you to manage the flow, see history, test and deploy.

Publish Flow

Create a new version of this flow in your workspace. Each change to the flow is temporarily autosaved in your specific session. Publishing the flow saves your local changes as a new permanent version anyone in your workspace can use. If you leave the editor without publishing, all your changes will be lost.

Run Flow

Run flow allows you to test your chatbot while you’re working on it, but also as a way to interact with finished flows. The button brings up a chat window. All chats get saved in history and you also get to see the past sessions with this flow:

CRON jobs

Set up recurring time-based flow executions. Let’s say you want a personalized weekly digest of news in your industry. You can create a flow that scrapes the internet or specific news portals, and then set up a CRON job to send it to your email each monday at 9AM.

History

Allows you to see and manage previous versions of your flow. This helps you and your teammates understand how the flow evolved, who is responsible and revert versions if necessary.

Flow management settings

If you hover over the Flow’s name in the top-left corner, a flow management menu appears with quick actions:

• Import and Export: Download and upload flows as a light JSON file.
• Edit details: Change the name and description of the flow.
• Templates: Import a template from the Flow Library.
• Share Flow: A shortcut to immediately deploy the flow as a chatbot.

Remember: Publish Flow does not mean you are publishing it publicly outside of your workspace. It only means you are saving a new permanent version within your workspace. If you leave the editor without publishing, all your changes will be lost.

AI Assistant

FlowHunt’s AI Assistant allows you to build flows by simply describing them in natural language.

Trigger the assistant by clicking the button in the bottom right corner. Then just describe the changes (or the entire flow) you’d like to make and wait for the assistant to do it’s magic.

Note: The assistant is a new feature in BETA mode and is liable to make mistakes or provide incomplete flows.

Components menu

Now on to the main star of the show. On the left side of the canvas, you’ll see the components button. Clicking it brings up the list of all the available components sorted into groups:

Adding and setting up components

To add a component, either click the plus sign next to it or simply click and drag-and-drop it:

Component Settings and Configuration

Nearly all components have their own additional settings that you can access by clicking on the component. These settings allow you to fine-tune how each component behaves within your workflow.

Most settings are optional, but they play a crucial role in ensuring efficiency and optimal performance by giving instructions, specifying scope or limiting usage.

Connecting components

Each component comes with appropriate input and output handles that allow it to be connected to other components. Some, such as Chat Input or Output will only have one set of handles, since they are designed to be at the start or at the end of flow.

All handles and components are color coded according to their categories, making it easier for you to intuitively match components to handles.

To connect components, click the handle and drag the line to the correct handle of the following (or previous) component:

Building a Flow

Let’s get familiar with the building process. We will create the simplest chatflow possible — a basic connection to an AI Agent powered by ChatGPT.

1. When you open a blank flow, the Chat Input and Chat Output components should already be present. If not, you can find them in the components list (in Inputs and Outputs sections respectively) and drag them in.
2. Go to the Agents category of components and pick AI Agent:
3. Connect the three via the green message/text handles. This means that the input/output processed here will be a plain text message:
   
4. You can technically start chatting now. But to get a better experience, try setting up the AI Agent .

That’s it! You’ve built the simplest possible chatflow in FlowHunt. But keep in mind that this flow is very basic — it has no tools to use, no history of past messages and no other source of data than what ChatGPT was trained on.

Next Steps

This chatflow is far from complete. Next you can try these:

• Add Chat History to make sure the chatbot remembers previous messages.
• Add Retrievers as tools to ensure the chatbot has up to date information with clear sources.
• Switch the LLM to one by a different provider (e.g. Claude, Gemini, xAI) by connecting an LLM component.

Note: Learn more about all available components in this section .

Remember: While chatbots and chat-based AI Tools are by far the most popular use cases, not all flows in FlowHunt are chatflows. You can build fully automated workflows that run on a schedule without your intervention. For example, to update a google sheet based on latest data from your CRM or to send a generated article right to your CMS.

Testing the Flow

Once you create the Flow, you can test it within the editor. Click the “Run flow” button in the top right corner and chat away.

Once you’re done building, you can either save your chatbot or tool locally by clicking “publish flow” or make it public by turning it into a chatbot with the “share” button.