Automating Cloud Operations Support

Aderant, a global provider of business management software for the legal industry, has implemented an intelligent ticket triage system to streamline its cloud operations support workflow. The company utilized Amazon Nova Lite through Amazon Bedrock to build a solution that automates context gathering, classification, routing, and knowledge enrichment for incoming support requests.

The Intelligent Ticket Analyzer supports Aderant's 38-person SierraOps team, which manages Expert Sierra across 268 client environments globally. Operating during scheduled hourly processing cycles on business days, the analyzer reviews newly submitted, unassigned tickets. It gathers operational context from multiple sources, including Jira, Confluence, Microsoft SharePoint, and Amazon Athena, to recommend team assignments and resolution starting points before automating approved routing and communication actions.

Addressing Manual Triage Challenges

Prior to implementing the automated system, the SierraOps team processed an average of 34 to 40 support tickets per week, with each ticket demanding 15 to 25 minutes of manual investigation before resolution work could begin. Engineers previously had to interpret incoming requests, locate client environment details, determine appropriate ownership, search documentation and ticket history, and handle assignment or redirection manually.

This manual process created operational bottlenecks. Misrouted tickets could languish in incorrect queues until discovered, and less-experienced team members often required additional time or guidance from senior engineers to gather context and make proper routing decisions. Aderant sought a repeatable investigative workflow to reduce this operational toil while preserving human oversight for lower-confidence decisions.

Serverless Workflow and Architecture

The Intelligent Ticket Analyzer operates as a serverless workflow orchestrated by an AWS Lambda function and triggered hourly on weekdays by Amazon EventBridge. For each unassigned ticket in the CloudOps queue, the system executes a five-stage process: identification, enrichment, classification, action, and observation.

During the classification stage, the ticket and its assembled context are sent to Amazon Nova Lite via the Amazon Bedrock Converse API for structured classification and recommended next steps. The architecture also incorporates Amazon CloudWatch for operational monitoring, while routing lower-confidence classifications to human review.

Model Selection and Evaluation

Aderant selected Amazon Nova Lite after evaluating multiple foundation models through Amazon Bedrock using real ticket data. Three primary factors influenced the choice: operational specificity in extracting resolution steps and correlating past issues, native AWS integration within existing security and SDK environments, and cost-effectiveness at scale.

This deployment builds upon Aderant’s earlier AI journey, extending past initiatives in unified search and on-demand investigation to scheduled, autonomous operational workflows powered by advanced foundation models.

Early Production Results and Efficiency Gains

During the first two and a half weeks of production between June 30 and July 17, 2026, the analyzer reviewed 109 tickets and achieved an estimated 96% routing accuracy. Based on previous pre-automation baselines, Aderant estimates that the system recovers between 8 and 14 engineering hours per week.

The total operating cost for the system remained under $30 per month, with Bedrock inference costs staying below $1 per month. Aderant notes that the analyzer processes internal operational ticket data and internal knowledge sources only, without accessing or storing client matter or application business data.

Sources

• AWS Machine Learning BlogAderant builds intelligent ticket triage with Amazon Nova