--- sourceDocument: Asset Management sourceDocumentLink: https://www.servicenow.com/docs/r/it-asset-management Release : - brazil ft:locale : - en-US ft:publication_title : - Asset Management ft:clusterId : - itam bundleId : - itam workflow : - Technology --- # AI-powered software spend detection # AI-powered software spend detection {#ariaid-title1} Release version: Brazil Updated March 12, 2026 ![](https://www.servicenow.com/docs/portal-asset/ico-clock) 3 minutes to read AI-powered spend detection identifies software purchases in imported spend transactions and normalizes the derived raw publisher and raw product to existing publisher and product records in Software Asset Management Content Library. This automation improves spend reporting accuracy and reduces manual classification effort. When the ServiceNow Otto for Software Asset Management (SAM) store app (sn_now_assist_sam) is installed, imported spend transactions are processed by two AI skills: the Spend transaction software classification skill and the Spend transaction software normalization skill. These AI skills improves classification accuracy for new software products and publishers, then links each software transaction to existing publisher and product records.{#spend-detection-ai-enhancements__spend-detection-ai-enhancements-para-1} ## How the AI-powered pipeline works {#spend-detection-ai-enhancements__ai-pipeline} After a spend transaction import completes, the SAM - Label Spend Transactions scheduled job runs the imported transactions through the following AI skills:{#spend-detection-ai-enhancements__spend-detection-ai-enhancements-para-2} Spend transaction software classification skill : Analyzes the vendor name, description, and GL account for each transaction to identify whether it represents a software purchase. For transactions identified as software, the skill also extracts the publisher and product. Spend transaction software normalization skill : Matches the extracted publisher and product to the existing records in the Software Asset Management Content Library. This linking supports accurate spend reporting and license position tracking. For more details, see [Classify and normalize software spend transactions by using ServiceNow Otto for Software Asset Management (SAM)](https://www.servicenow.com/docs/_2lJJrBsow0YVUYrFYUUSA "Classify imported spend transactions to identify software purchases and normalize the derived raw publisher and raw product by matching them to existing publisher and product records. This automation reduces manual effort and improves spend reporting accuracy.") and [Skill inputs and triggers for ServiceNow Otto for Software Asset Management (SAM)](https://www.servicenow.com/docs/gByOm~d44ygeT0vMJKuvSQ "Get a quick overview of the skill inputs and triggers for ServiceNow Otto for Software Asset Management (SAM). By configuring the inputs or triggers for a skill, you can determine how and when a skill is used."). Transactions not identified as software are not processed by the Spend transaction software normalization skill. Note: If the ServiceNow Otto for Software Asset Management (SAM) store app isn't installed or either AI skill is inactive, the SAM - Label Spend Transactions job uses the existing machine learning pipeline instead. ## How transactions move through the labeling pipeline {#spend-detection-ai-enhancements__state_lifecycle} The SAM - Label Spend Transactions scheduled job updates the State, Is software, and Is managed fields on the Software Spend Transactions \[sam_spend_transaction\] table for each transaction based on the outcome of the classification and normalization skills. The State field reflects the pipeline outcome for each transaction: New : Default state when a transaction is imported or manually created. A transaction stays in New if the pipeline exhausts the maximum number of retry attempts without reaching a terminal state. Labeled : The classification skill identified the transaction as a software purchase, and the normalization skill matched both the publisher and product to existing records in the Software Asset Management Content Library. Partially labeled : The classification skill identified the transaction as a software purchase, and the normalization skill matched the publisher, but not the product, to an existing record in the Software Asset Management Content Library. The Is software field is set to true. Unlabeled : The pipeline couldn't classify or normalize the transaction after the maximum number of retry attempts. Manually labeled : A user manually set or updated the Is software, Publisher, or Product fields on the transaction. When the classification skill identifies a transaction as a non-software purchase, the Is software field is set to false, and the transaction bypasses the normalization skill. After a transaction reaches the Labeled or Partially labeled state, the Is managed field is evaluated. When a matching software model exists in your Software Asset Management Content Library for the normalized publisher and product, the Is managed field is set to true. The maximum number of retry attempts is controlled by the samp.spend_prediction_service.max_label_attempts system property, which defaults to 3. You can update this property at any time. For more details, see [Software Asset Management properties](https://www.servicenow.com/docs/6ydCjyeBRTYF3NVanLG~Xg "You can set default reconciliation properties such as grouping and reconciliation debugging."). ## AI activity log {#spend-detection-ai-enhancements__ai_activity_log} Track how many software transactions are processed by the AI pipeline using the AI activity log card. This card shows a rolling count of transactions from the Software Spend Transactions \[sam_spend_transaction\] table where the prediction method is AI, the transaction is a software purchase, and the state is Labeled or Partially labeled. To access the card, go to Software Asset WorkspaceSoftware asset overview. The card appears under the Activity center section. **Related concepts** * [Managing software spend in the Software Asset Workspace](https://www.servicenow.com/docs/5q5H35Rqv3BDKdh185ERoQ "The Software Spend Detection feature is available in the Software Asset Workspace under License operations. Use Software Spend Detection in the Software Asset Workspace to import financial transactions, review labeled transactions, and track software spending across your organization.") * [AI in Software Asset Management](https://www.servicenow.com/docs/tgJzs7A1uoGCi5Kth6uzgg "ServiceNow Otto for Software Asset Management (SAM) uses AI to automate software asset management operations, improve compliance, and optimize license management.") * [Generative AI skills in ServiceNow Otto for Software Asset Management (SAM)](https://www.servicenow.com/docs/OivvZCHlO91n6dRc44CO9Q "ServiceNow Otto for Software Asset Management (SAM) includes generative AI capabilities that help ServiceNow Otto for SAM managers streamline daily workflows, gain real-time insights, and automate repetitive tasks throughout the software asset lifecycle.") **Related reference** * [Skill inputs and triggers for ServiceNow Otto for Software Asset Management (SAM)](https://www.servicenow.com/docs/gByOm~d44ygeT0vMJKuvSQ "Get a quick overview of the skill inputs and triggers for ServiceNow Otto for Software Asset Management (SAM). By configuring the inputs or triggers for a skill, you can determine how and when a skill is used.") *[\>]: and then