KnowledgeForge: mining gold from the ITSM ticket graveyard

The KnowledgeForge solution addresses the issue of valuable but unused knowledge locked away in resolved IT Service Management (ITSM) tickets. When IT support teams resolve tickets, they often contain useful information, such as symptoms, root causes, and fixes applied by engineers. However, this knowledge remains hidden inside ticket history, making it inaccessible for future engineers facing similar problems.

At the same time, the knowledge base itself faces challenges, such as duplicate articles, outdated content, and varying quality. To tackle these problems, KnowledgeForge works on both ends of the knowledge gap.

The system mines resolved incident tickets for new articles while simultaneously improving the existing knowledge base by sorting articles by type, removing duplicates, assessing their quality, and refining weak content. A knowledge manager reviews and approves the final results, ensuring that only high-quality content goes live. This process helps create a more effective and efficient knowledge base.

The core components powering KnowledgeForge include Amazon Bedrock for generation and content improvement, Amazon S3 Vectors for duplicate detection, and AWS Step Functions for workflow orchestration. By combining these AWS services, the solution provides a scalable and efficient way to process large amounts of document data using generative AI.

If you are building a large-scale document-processing pipeline with generative AI, you can leverage these same patterns to create a more efficient and valuable knowledge base for your organization.