​
Creating an agent workflow

Agent workflows allow you to connect multiple AI agents and tools to create powerful automations. Use our simple drag and drop editor to design and deploy your custom workflow in minutes.

To create a new agent workflow:

  1. Sign Up in GenFuse AI
  2. Once you’ve logged in, go to the Agent Workflows page (in the sidebar)
  3. Create a new workflow by clicking the button in the top right corner.

​
Customizing the agent workflow

​
Add a Trigger

Triggers are events that start your workflow. Examples include manual triggers (run on demand), scheduled triggers (e.g. every weekday at 7 AM), or event-based triggers (e.g. when an email is received).

In this example, we’ll set up a manual trigger that takes a company website as input.

  • Click the Trigger node in your workflow
  • Select Manual Trigger
  • Click Add Input and add Company Website as an input — this lets you provide custom data each time the workflow runs.

​
Add an AI Agent

AI agents are intelligent systems powered by LLMs that can perform multi-step tasks without human intervention. They can be equipped with tools and can autonomously decide how to use them to complete tasks.

  • Click the + button to add a new step.
  • Select Agents
  • Click New Agent to create an agent.

​
Build the Agent with AI

Describe the task you want your agent to perform, and let AI build a powerful agent for you. It will set up the agent’s instructions and add the necessary tools based on your task. You can then customize the agent to fit your needs.

  • Click on the added AI Agent node
  • Click the Create with AI button
  • Set the agent description to:
text
Given a company website, build a company research agent that scrapes the website and uses Google Search to generate a concise, accurate executive report in HTML format. 
The report should contain the following sections: Executive summary, key product or service offerings, customer case studies or testimonials (if available).
  • Click Build Agent
  • Once your agent is built, accept it and make any necessary changes. You can customize the agent instructions, add or remove tools or change the underlying LLM model (under Settings).

​
Add Agent Input

Each agent can accept one or more inputs. By configuring these inputs, you can pass data from earlier steps to the AI agent. Use the + button in the text box to insert output from any previous step into the current step.

Optionally, you can reference any agent input within the agent task.

  • If it’s not already added, create an agent input named Company Website.
  • Click the + button to map the company website from the Manual Trigger step to the value of this input.

​
Add Tools

Each agent can be equipped with tools. Based on the task, the agent will use them as needed to complete the work.

  • Add or remove tools as required. For this example, you can add the Scrape Website and Google Search tools.

​
Update LLM Model

Each agent is powered by an LLM model, which you can choose under Settings. Different agents in a workflow can use different LLM models.

  • You can change the LLM model under settings.

​
Add Email Integration

App integrations enable you to connect to third-party services like Google Search, Sheets, Email, Linkedin, Web Scrapers and more.

  • Click the + button to add a new step.
  • Select App Integrations
  • Click Notification and select the Send Email Notification action.
  • Configure the Send Email Notification node by clicking on it
  • Set the email Message to the report generated by the Company Research Agent by clicking the + button
  • Add your email id in the To field and add an optional Subject

​
Run the agent workflow

  • Save the workflow
  • Test the workflow by providing the company website you want to research
  • Once the run completes, you’ll be able to see the exact steps that were taken to complete the task. And you should’ve received your first email from the AI agent!
  • Customize the workflow as needed
  • Run your workflow easily and see all past runs in the Use tab

​
Share your workflow

Congratulations! You’ve successfully built your first agent workflow.

Share it with your team by clicking the Share button. Your teammates can now clone and use this workflow.

Was this page helpful?

Suggest editsRaise issue