Kernel: Internal IT Support with Azure OpenAI
Client
Kernel is the world’s leading producer and exporter of sunflower oil, the largest producer and exporter of grains in Ukraine, and a major supplier of agricultural products from the Black Sea region to international markets. The company accounts for about 8% of the world’s sunflower oil exports. Kernel delivers its products to over 70 countries. Since November 2007, the Company’s shares have been traded on the Warsaw Stock Exchange (WSE).
Challenge
Kernel aimed to improve the customer support experience for their employees by providing faster, more accurate responses to inquiries.
Integrating new automation tools with the existing IT support infrastructure was complex: the AI-based system had to interact seamlessly with Jira, Confluence, and SharePoint Online without disrupting existing workflows. The system also had to handle a wide range of inquiries — from simple questions to complex issues requiring detailed responses — while keeping answers accurate and up to date, which meant regularly indexing content from all connected knowledge sources.
Solution
DevRain implemented an AI-based system enhanced with agentic Retrieval Augmented Generation (RAG) to leverage their existing knowledge bases.

The system operates through two primary user flows. When an employee creates a support request, the AI analyzes its content in real time and, using the knowledge base built from Confluence, Jira, and SharePoint Online, immediately suggests the essence of the request and possible solutions — helping employees find answers quickly without waiting for an engineer.
When a support engineer opens a ticket, the AI has already prepared a draft response or advice on resolving the issue, based on similar past requests and relevant knowledge base content. This gives engineers a head start and lets them focus on complex tasks that require human judgment.

Key Components:
- Azure OpenAI: Advanced language model for natural conversation
- Agentic RAG: Technique to enhance AI responses with the retrieved information
- Data Sources:
- Confluence: company internal knowledge base
- Jira: issue tracking and project management
- SharePoint Online: internal documents - technical manuals and guides
Implementation
- Data Indexing: content from Confluence, Jira, and documents was processed and indexed to AI knowledge base. Documents (PDF, Word etc) are processed with Azure Document Intelligence to retrieve content.
- RAG Integration: the system was set up to retrieve relevant information from AI knowledge base and use Azure OpenAI to answer user queries. The system provides citations so users can validate the correctness of the answers and check details if needed.
- Real-time updates from Jira, Confluence, and SharePoint are being processed, and AI knowledge base was updated.
JIRA and Confluence: REST API is used. To receive real-time updates - Webhooks.
SharePoint Online: Microsoft Graph API is used. To receive real-time updates - SharePoint Online Webhooks.
Conclusion
By combining Azure OpenAI with agentic RAG and leveraging existing knowledge bases, Kernel improved its customer support efficiency and effectiveness, improved employee and customer satisfaction, and reduced costs.
References
https://devrain.com/blog/crafting-technical-specifications-for-custom-chatgpt-implementations
Related cases
Enabling Quick Access to Corporate Knowledge with Agentic RAG using Azure OpenAI