Zia-powered Workflow Assist is a virtual workflow expert. It generates, optimizes, and summarizes workflows and suggests steps and actions to accelerate and automate workflow design.
Using Workflow Assist, you can build workflows quickly. It understands designs and inputs, and recommends the right steps, conditions, and logic to create workflows.
You can interact with Workflow Assist through:
Text commands
Canned prompts
Images (flowcharts, workflow images, etc)
AI Requirements for Workflow Assist
Workflow Assist responds only to workflow-related queries. Irrelevant questions are not supported.
It provides guidance based on the current module.
Example: While creating an Incident workflow, Assist responds only to incident-related queries.
On switching to the Changes module, it provides assistance relevant to changes.
Builds Workflows - Creates workflow instantly based on the requirement.
Provides Suggestions - Recommends the next steps after each stage to ensure workflow completeness.
Optimizes Workflows - Identifies redundant steps and removes them to improve workflow performance.
Helps Configure Conditions & Logic - Enables you to configure branching rules and conditions with simple prompts.
Automates Common Tasks - Creates notifications, timers, and approvals quickly with minimal input.
Troubleshoots Workflows - Locates nodes in the canvas, highlights unconnected nodes, and provides fixes.
Accesses Help Guide - Fetches help articles from the knowledge base for quick reference.
Go to Setup > Zia > Artificial Intelligence > GenAI Features.
Enable the toggle on the Workflow Assist card.
Select the required AI provider. Supported AI providers are Zia (Zoho-hosted LLM), ChatGPT, and Azure OpenAI.
On selecting Azure OpenAI or ChatGPT, you are prompted to review and accept the data privacy notice.
Finally, click Confirm.
Text-based workflow generation - Requires a highly capable model such as Qwen-2.5-coder-instruct-32b-zlabs, ChatGPT 4, or higher versions.
Image-based workflow generation - Requires a multi-modal model such as ChatGPT 4 or higher to handle flowcharts and graph-like images.
Workflow Summary: Supported across models that can process workflow data.
Go to the Workflow Assist card.
Click the three vertical dots and select Configure AI provider.
Select the required AI provider and click Confirm.
After enabling Workflow Assist, you can access it from the workflow editor page.
Go to the workflow editor page.
Click Ask Zia on the top-right of the page or click Use Ask Zia in the canvas.
Type your requirement in the chat box, or click Explore Sample Prompts to choose from available options.
Provide image input by clicking the plus icon in the chat box and uploading the image.
Image input is not supported for Zia (Zoho-hosted LLM).
After providing the input, respond to the follow-up questions to clarify your requirements. Assist will then provide an overview of your requirement and ask for confirmation before building the workflow.
Click Draw Workflow to carry out the workflow creation, or continue the conversation by providing instructions and additional details to decide how the workflow should be created.
Nodes created with assist may contain incomplete data. They are marked with an indicator icon 
for easy identification.
You can save an invalid workflow created with Assist and revisit it later to complete and make it ready for use.
Conversations are not stored. The chat will be cleared when the page is refreshed or closed.
Currently, Workflow Assist does not support User Transitions and creation of the following node types: Third Party Notifications, Resource Switch, Resource Condition, ZFAction, Action Sequence, and Zoho Circuit.
After creating the workflow, you can fix the incomplete nodes by using the following options:
Let's see each option in detail.
Automatically fills in details based on the workflow context. Use this option when you want the system to complete most of the information for you.
Hover over the incomplete node.
Click 
and select Auto fix.
You will be directed to the corresponding node form, pre-populated with data based on the workflow context.
Review the form and add or update the required fields.
Click Save or Save and Select. The node will be fixed automatically.
Helps fix invalid nodes and modify valid nodes with Zia. Use this option when you need assistance or are unsure about the required inputs.
Hover over the incomplete node.
Click 
and select Ask Zia.
The node title will appear as a prompt in the chat box. Press Enter or click the send icon 
.
Ask Zia provides the details required to complete the node and asks for your input.
Based on your response, it configures the node form and displays it for review.
Add or update the details as required, then click Save or Save and Select.
Generate a text-based summary of the graphical workflow.
Go to the workflow editor page.
Click 
on the top-right of the page and select Summarize.
Alternatively, you can generate a summary from the workflow list view. Click 
beside the required workflow and select Summarize.
After the summary is generated, you can copy or download it using the corresponding buttons at the bottom of the summary page.
Use Case 1: Creating a Change Workflow with Workflow Assist
Jessica, a Change Manager, often spent hours creating workflows by configuring conditions, actions, and logic. Despite the effort, the workflows sometimes failed, and she had to seek help from the support team.
With Workflow Assist, creating workflows became effortless. Now, Jessica opens the workflow editor and types: “Create a normal change workflow for Email Spam Filter Adjustment.” Within seconds, Assist drafts the workflow. She uses Auto Fix to resolve incomplete notification nodes and Ask Zia to get guidance on fixing condition nodes.
Jessica then reviews and saves a fully functional workflow, which is ready in minutes, saving hours of manual effort and reducing her dependency on the support team.
Use Case 2: Replicating Workflows with Image Input
Rachel, an HR Operations Manager, needed to create an employee onboarding workflow in the IT help desk instance. The same workflow already existed in the HR instance. Instead of rebuilding it from scratch, Rachel used the image upload option in Workflow Assist. She uploaded the image of the onboarding workflow from the HR instance.
Assist scanned the image, interpreted the steps, and asked a few clarifying questions to confirm the conditions, nodes, and logic. It then provided an overview of the requirement and asked for confirmation before building the workflow. After Rachel confirmed, Assist instantly drafted a workflow.
Rachel then used Auto Fix and Ask Zia to resolve the incomplete nodes. In a short span of time, the onboarding workflow was fully replicated and ready to use. This saved Rachel a significant amount of time and effort..
