- Staff turnover leads to information about important incidents being lost. When a new problem arises, its resolution starts from scratch and takes a lot of time.
- Information about products and infrastructure is scattered across different Jira documents, diagrams, and incidents. When analyzing a problem, it can be difficult for a specialist to collect the necessary information and build a complete picture of the dependencies of components.
- It also happens that among many similar incidents it is difficult to find the one that is most relevant to the current problem and can help in solving it.
Technical support staff spend most of their time on:
- Searching for information about the architecture and dependencies of components
- Determining the source of the problem
- Searching for ways to solve it
- Restoring the operation of all dependent components of the failure.
Increasing the availability of information can significantly reduce the response time to incidents, which means reducing system downtime and associated losses.
We propose to build a single technical support knowledge base that will combine information from:
- Technical documentation and FAQ,
- Jira or other bug trackers,
- E-mail correspondence and work chats.
Knowledge elements in the base are linked to products, functional modules, types and levels of problems, responsible persons, etc. The knowledge base is constantly updated by collecting information from all sources.
- A support agent briefly describes the problem, or the assistant receives the problem description from a new Jira ticket or email
- The assistant finds relevant tickets and other knowledge base elements
- If suitable information is found in the FAQ or documentation, the assistant summarizes the information and provides recommendations
- The assistant then analyzes similar incidents and provides a brief summary for each of them with a link to the source
The intelligent assistant is based on the DataVera EKG Language Processing product. It uses NLU and LLM for text processing, the GraphRAG approach for knowledge base formation.