{"revision":"c10276e953456835cdbb1f3a60d58ae1980be353","document":{"id":23002,"path":"intelligent-experiences/classify-tasks.md","title":"Platform Classify tasks agentic workflow","bundle":null,"doc_type":null,"release":"australia","body":"\n# Platform Classify tasks agentic workflow\n\nUse the Platform Classify tasks agentic workflow to gather relevant information about tasks automatically and make decisions about priorities and assignments.\n\n## Classify tasks overview\n\nThe Classify tasks agentic workflow can help improve efficiency and accuracy by automatically gathering information, prioritizing tasks, assigning teams, detecting sentiment, and generating task summaries.\n\nThe agents, tools, and triggers that are associated with the Classify tasks agentic workflow are provided by AI applications. You can [activate the agentic workflow template](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/activate-aia-use-case.md) by making triggers active and setting the display settings to include the ServiceNow Otto panel. If you want to change this agentic workflow's instructions, you must [duplicate it](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/clone-aia-usecase.md), adjust the settings to suit your specific needs, and activate the duplicated version of the agentic workflow instead.\n\n**Note:** Depending on your license, you will have access to certain application features, generative AI skills, agentic workflows, and AI agents. For more information, see [ServiceNow product tiers](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/ai-native-sku-overview.md).\n\n## Prerequisites and setup\n\nTo access this workflow, you must have ServiceNow Otto for Platform installed on your instance, which you can get if you install any other AI application.\n\nFor this agentic workflow to behave as expected, you must also configure Group Action Framework \\(GAF\\). See [Set up AI Search for Group Action Framework](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/setup-ai-search-gaf.md) and [Configure Group Action Framework](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/configure-gaf.md) for more information on getting started with GAF.\n\n## Role filtering\n\nRequired role: sn\\_uxc\\_gen\\_ai.platform\\_ai\\_classify\\_tasks\n\nAgentic workflows and their AI agents use [role masking](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/implement-aias-security-new.md) to determine which users can access them. Ones installed with your applications have specific roles that come included with the application. If you select **Users with specific roles** for user access, you must configure the security controls to include these roles. For the instructions to change the security controls, see [Define security controls for an agentic workflow](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/define-sec-controls-aw.md).\n\nIn the data access settings, you must also add the necessary roles to enable reading of the tables for the records you want to classify. For example, you can add the itil role to the agentic workflow's list of approved roles so that it can access Incident records.\n\n## Additional configuration\n\nYou can change different settings related to the agentic workflow by changing values for the Now Assist Skill Config Var Set. To access the variable set and make changes, do the following while in the Platform AI Agents and Skills scope:\n\n- Go to the Now Assist Skill Config \\[sn\\_nowassist\\_skill\\_config\\] table.\n- Open the record named **Task Classify Tasks Skill Config**.\n- In the Now Assist Skill Config Var Set related list, select **Field Predictor**.\n- Edit the variable values.\n- Save or update the record.\n\nThe Classify tasks configuration variable set includes the following variables. You can configure either the AIS fields or the GAF field for determining how the agentic workflow gathers what work the user has. If you configure both, GAF takes priority when running the agentic workflow. For more information about GAF, see [Group Action Framework](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/group-action-framework.md).\n\n\n
\n\nConfig field\n\n\n\nDescription\n\n
\n\nSearch Mode\n\n\n\nOptions include Keyword, Semantic, or Hybrid. Keyword-based searches look at individual phrases, while semantic-based searches rely on the phrasing of utterances as well. Hybrid utilizes both strategies.\n\n Default: Keyword\n\n
\n\nAIS Search Fields\n\n\n\nFields used by AI Search operations to find similar records for improved prediction context\n\n
\n\nCapability\n\n\n\nOneExtend capability for enhanced field prediction functionality. See [Generative AI Controller reference](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/generative-ai-controller/reference-for-generative-ai-controller.md) for more details.\n\n
\n\nGAF Config\n\n\n\nSkill configuration for [Group AI Framework \\(GAF\\)](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/group-action-framework.md).\n\n
\n\nAIS Return Fields\n\n\n\nFields returned by AI Search to the agentic workflow to base decisions on\n\n
\n\nGAF Config\n\n\n\nGroup Action Framework grouping configuration record, which is a collection of groups of records to make searching easier\n\n
\n\nAIS Search Profile\n\n\n\nProfile for AI search, such as ServiceNow Otto for Virtual Agent.\n\n
\n\nCapability Payload\n\n\n\nServer-side script to generate custom payloads for OneExtend capability integration\n\n
\n\nAIS Fallback Table\n\n\n\nTable used if GAF is not configured\n\n
\n\nReturn Fields\n\n\n\nField values returned from identified relevant records\n\n
\n\nCheck for Issue Readiness\n\n\n\nUses the information provided by the Issue Readiness AI agent\n\n **Warning:** Only enable if the Issue Readiness AI agent is active on your instance. Otherwise, the agentic workflow will not execute.\n\n
\n\nFields\n\n\n\nField values to predict and update\n\n
\n\nTable\n\n\n\nTable where you want to execute the agentic workflow on\n\n
\n\nSearch Term Source\n\n\n\nOptions include Record-based and Utterance-based. Record-based derives search terms for generating context from the record itself. Utterance-based relies on the user-submitted utterance for search term options.\n\n Default: Record-based\n\n
## Accessing the Classify tasks agentic workflow\n\nTo access the agentic workflow:\n\n1. Navigate to **All** > **AI Agent Studio** > **Create and manage**.\n2. Select **Classify tasks**.\n\nThe first step of the guided setup includes a complete list of included AI agents. Selecting the name of an AI agent opens it in a new browser tab, where you can see the full description, role, list of steps, and tools. Tools are displayed in the second step of the AI agent guided setup, Add tools and information.\n\n## In-product agentic AI and UI actions\n\nAgentic workflows can be accessed in the Core UI and in workspaces in the AI Activity panel. From there, you can track their progress, provide or review input, and see the results of the work performed. For more information, see [In-product agentic AI](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/in-product-agentic-ai.md) for more details about the AI Activity panel.\n\nTo enable users to access agentic workflows with UI actions, you can open the agentic workflow in AI Agent Studio and navigate to the **Select channels and access** step. You can select a UI action as a possible way to access the workflow\n\nIf you don't see your UI actions after configuring it in AI Agent Studio, ensure that the property **com.glide.agentic\\_processes\\_view.enabled** is set to `true`. See [Enable the in-product experience for agentic workflows](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/enable-inproduct-aia.md).\n\n## Testing the Classify tasks agentic workflow\n\nYou can manually test an agentic workflow execution or access on the Testing page of AI Agent Studio if you have the sn.aia\\_admin role and all other roles configured [in the security controls](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/define-sec-controls-aw.md). Start a manual test, select a test type and the name of the workflow, and use utterances in the Task field like the following samples. See [Test an agentic workflow execution](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/test-aia-use-case.md).\n\nIf you want to evaluate the agentic workflow over many different execution logs, run an [automated evaluation](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/execute-aia-eval.md).\n\n## Sample utterance\n\nAfter the workflow has been activated in AI Agent Studio, enter these or similar phrases in the ServiceNow Otto panel to trigger the workflow. You must have the sn.now\\_assist\\_panel\\_user role to run the workflow. 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.\n\n- Populate the priority and assignment group for INC0001\n- Populate the empty fields on INC0001\n- Find the priority for INC001\n\n## AI agents used in the Classify tasks agentic workflow\n\nThe following table lists the agents that are used in the Classify tasks agentic workflow.\n\n**Important:** In the Define availability step of each AI agent's guided setup, make sure that the Status toggle is enabled to activate the AI agent.\n\n|AI agent name|AI agent description|Role required|\n|-------------|--------------------|-------------|\n|Record field value prediction AI agent|Fetches, predicts, creates, and updates records with the provided details. Provides summary and justification for recommendations.|sn\\_uxc\\_gen\\_ai.platform\\_ai\\_field\\_predictor|\n\n## Other Platform agentic workflows\n\nFor more information on other agentic workflows that are associated with the Platform workflow, see [Platform agentic workflows](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/intelligent-experiences/platform-use-cases.md).\n\n"}}