Aderant automated support ticket classification with the Amazon Nova Lite model
Aderant deployed automated support ticket classification using the Amazon Nova Lite model through Amazon Bedrock. Over 2.5 weeks of operation, it achieved, according to the company, 96% routing accuracy and estimated savings of 8–14 hours of engineering time per week at costs below 30 USD per month.
Aderant, a provider of law firm management software, deployed Intelligent Ticket Analyzer to automatically classify support tickets for its 38-member SierraOps team, which serves 268 client environments. The system runs as a serverless workflow controlled by a single AWS Lambda function, triggered hourly on weekdays through Amazon EventBridge. For each unassigned ticket, it gathers context from Jira, Confluence, Amazon Athena and Microsoft SharePoint and uses the Amazon Nova Lite model through Amazon Bedrock to suggest a team assignment and a starting point for resolution; when confidence is sufficient, it can route the ticket autonomously, otherwise it passes the ticket on for human review.
According to Aderant, during the first 2.5 weeks of operation (30 June to 17 July 2026), the system processed 109 tickets with routing accuracy of approximately 96 % (4 routing errors). Before automation, manual triage took 15–25 minutes per ticket at a volume of 34–40 tickets per week; the company estimates savings of 8–14 hours of engineering time per week (32–56 hours per month). According to the company, total operating costs for the system are less than 30 USD per month, of which inference costs through Amazon Bedrock are under 1 USD per month.
According to the company, the Amazon Nova Lite model was selected after comparing several foundation models using real ticket data — it excelled at extracting specific resolution steps and connecting historical issues, integrates natively with the AWS environment (IAM, SDK, Bedrock Converse API), and its cost profile made it possible to analyze regular tickets as well, rather than only priority tickets. The company emphasizes that these results come from the initial production period, rather than a long-term benchmark, and that the system does not work with client data or business data from client applications, only internal operational metadata.
Why it matters
The case shows that even a smaller foundation model deployed through a serverless architecture can deliver measurable time savings for recurring administrative work at negligible operating costs. For companies with a similar volume of internal tickets (dozens per week), it provides a concrete example of how to automate the first stage of triage without having to replace human decision-making — cases with low confidence remain with a human.
Relevant practical impact
What this means
For a business
Companies running internal support/helpdesk services can use a combination of a foundation model and serverless orchestration (Lambda, EventBridge) to automate routine ticket classification at low operating costs, freeing up experienced engineers for more complex work.
ProcessesCheck the original
Event sources
only one source so far · 1 publisher, 0 independent. We count feeds from the same owner only once.