Overview:
This ServiceNow’s out-of-the-box Agentic Workflow streamlines task processing and issue reporting by leveraging AI-powered image analysis. Instead of manually filling out forms, users can simply upload a photo—such as a broken chair, a malfunctioning laptop, or a compliance concern. The AI analyzes the image, identifies the issue, and automatically creates a well-categorized incident with the appropriate priority and description. The workflow then initiates key resolution steps including triage, planning, and closure, reducing manual effort and saving time. Users can begin in supervised mode to review and approve details or enable autonomous mode for full automation with a simple configuration change.
Find more details in our product documents - Platform Process images for new tasks agentic workflow
What Problem Is It Solving?
Organizations often struggle to efficiently process image-based requests, plan work, and ensure teams are prepared before execution. Without the Image Process AI Agents, employees must manually create tasks and enter all details themselves. Supporting images are rarely attached because there's no automation to interpret them.
This results in:
- Slower task creation
- Higher risk of missing critical information
- Reduced accuracy and productivity
Examples:
- Frictionless Issue Reporting
Employees can simply upload a photo of a damaged device, eliminating the need to fill out complex forms or type detailed issue descriptions. - Accurate Issue Identification
The AI Agent analyzes the uploaded image, automatically detecting device type, damage severity, and relevant context to ensure precise ticket creation. - Streamlined Incident Creation
AI populates all necessary incident fields, assigns the correct support group, and sets priority—minimizing manual effort and errors. - User Confirmation & Control
The system presents the extracted details back to the user for review and approval, ensuring transparency and user confidence. - Faster Resolution Start
Quick, accurate ticket creation accelerates issue tracking and assignment, reducing resolution time. - Improved Service Experience
This seamless, image-driven process enhances employee satisfaction by making IT support requests intuitive and efficient.
Setup Requirements:
- License required – Professional or Enterprise+
- Latest Version - Minimum patch Yokohama Patch 10 or Zurich Patch 4+
(Note: If you are in older version (YP6 /ZP1+) you may see some changes compared to latest versions as we have enhanced the workflow.)
- Plugins: Must have Now Assist for Platform (included with apps like Now Assist for ITSM, CSM).
- Below screen shot is an example for Now assist for ITSM which has also activated dependency plugin Now Assist for Platform.
- Required User Roles:
- sn_uxc_gen_ai.platform_ai_image_processor role is required to allow users upload image.
- Invoking users may have more access than an AI agent/agentic workflow needs, so limit security risk by assigning only specific approved roles.
The agent will only access data available in the overlap between the user’s roles and the approved roles (e.g., choosing itil prevents admin‑level access).
To give broader access than the invoking user, run the AI agent/agentic workflow as an AI user.- You can also run this workflow on the Testing page of AI Agent Studio with the same utterance in the Task field if you have the sn.aia_admin role.
- Activation:
To activate the workflow and AI agents by enabling trigger from the Now Assist virtual agent or Now assist panel, and assign the necessary roles for access.
- Activating Agentic workflow:
Go to All -> AI Agent Studio -> Create and Manage -> search for ‘Process Images for new Tasks’ -> click on it and go to ‘Define key requirements’ and scroll down to go to AI agents. The workflow involves two AI agents.
To configure security controls for user and data access for the workflow, please proceed as follows:
- Click 'Define user access' and add the role 'sn_uxc_gen_ai.platform_ai_image_processor'. If you require the workflow to create incidents from images, appropriate ITSM roles must be granted to enable incident creation in the incident table. In this context, the 'ITIL' role has been added to facilitate incident creation.
- Similarly, if the workflow is expected to create cases or HR tasks, relevant business unit-specific roles are necessary to ensure the workflow has adequate permissions to create records in the respective task tables.
- Include roles for 'Define data access' as well.
- To launch the agentic AI experience based on your chosen conditions or schedule, set trigger conditions in the ‘Add triggers’ section.
- Enable agentic workflow triggers in the Now Assist Panel or in-product experience for UI actions on a specific record. Activate this feature in the 'Select channels and status' section.
Activating workflow AI agents:
- Return to ‘Define Key Requirements’ activate each AI agent for the Virtual Agent assistants so Now Assist Virtual Agent can trigger them and return results to users in portal.
- configure security controls for user and data access for the AI agents, please proceed as follows:
- Click 'Define user access' and add the role 'sn_uxc_gen_ai.platform_ai_image_processor'. If you require the workflow AI agent to create incidents from images, appropriate ITSM roles must be granted to enable incident creation in the incident table. In this context, the 'ITIL' role has been added to facilitate incident creation.
- Similarly, if the workflow AI agent is expected to create cases or HR tasks, relevant business unit-specific roles are necessary to ensure the workflow has adequate permissions to create records in the respective task tables.
- Include roles for 'Define data access' as well.
- Go to ‘Select channels and status’, and enable ‘Engage via Virtual agent assistants’ This lets users upload images and create tasks using Now Assist virtual agent chat; ensure virtual agent access is active for portals.
Choose chat assistants-> Select drop down and choose Now Assist in Virtual Agent and make display active.
Note: If you have configured other Now Assist in Virtual Agent other than default, you can consider but make sure you have them activated in the portal where you are making Now Assist Virtual Agent to be activated for workflow AI agents to get trigger for the images.
- Scroll down -> select active for ‘Status’-> and then click on save and test.
**Repeat the same steps for other AI agent - Document and visual insights AI agent
Note: Make sure you are in right application ‘Platform AI Agents and Skills’ when saving changes.
Sample Utterances or Prompts:
“Create a task from an image."
"Analyze a photo and open an incident."
"Use an image to log a service request."
How Does It Work?
This agentic workflow uses below AI agents:
- Image Processor Agent
- Document and visual insights
Note: Earlier versions like YP6/ZP1 may differ from the latest release, as we've unified our AI agents to enhance performance and speed. Upgrading ensures you access all new features and improvements.
- When an utterance is given, Now Assist Virtual Agent or Now Assist Panel identifies the relevant workflow, activates the Agentic workflow orchestrator, and triggers AI agents to carry out the next steps.
- The screenshot below shows the utterance ‘create a task from an image’ sent by user in the Now assist virtual agent chat. This triggers the Process images for new tasks Agentic workflow orchestrator to activate the Image Processor Agent AI agent, which requests the user to upload an image.
- After uploading, a unique file ID is generated and passed back to the orchestrator. The orchestrator then calls the Document and Visual Insights AI agent to analyze the image and extract its information, as well as suggest relevant knowledge base articles or Q&As.
- If the user still wants to create a task based on the extracted response, the AI agent transfers control to the orchestrator, which prompts the Image Processor Agent to identify the correct task table and create a preview for user edits. Upon user approval, a new task record is created, and the task number is provided.
A new task record may resemble like below example incident record.
Demo
Outcomes Benefits
- Faster Task Creation - Customers can create tasks instantly by uploading an image, eliminating slow manual data entry. This dramatically speeds up the process of reporting issues or incidents.
- Higher Accuracy and Fewer Mistakes - AI‑generated structured data ensures more accurate task information, lowering the risk of incomplete or incorrect submissions.
- Clear and Guided Submission Experience - Customers can review and adjust the AI‑extracted details before submitting, ensuring clarity and preventing unnecessary back‑and‑forth with support teams.
- Faster Resolution & Improved SLAs - Because tasks are more accurate and better structured from the start, support teams can resolve issues more quickly—directly improving customer outcomes.
Suitable Personas:
Employees or authorized end users with image upload roles assigned.
Limitations:
Current image-to-task workflow supports only certain file types Jpeg, pmg for now
For more information:
- Product documentation - Platform Process images for new tasks agentic workflow
- Explore platform product resources here, and review the list of helpful resources under Platform Agentic Workflows and AI Agents.
