Challenge
Make HR knowledge accessible through the questions employees actually ask, while allowing administrators to add content, organize categories and choose when to sync it into the search system.
AI & Knowledge Management
HR Knowledge Chatbot
A conversational HR knowledge web application that connects an administrator-managed knowledge base with Azure AI Search and Azure OpenAI. Context-aware query rewriting supports follow-up questions, while an AI workflow coordinates retrieval, answers and relevant attachments.
Approach
Store concise knowledge entries focused on individual topics, organized by category, with one attachment per entry. Administrators add content and select Sync to update the search index. The Conversational RAG workflow uses the previous six conversation items to reinterpret the current question before retrieval. Azure AI Search retrieves relevant content using semantic search and embeddings, and Azure OpenAI composes an answer grounded in that content. An AI decision step determines whether an attachment from the retrieved knowledge is needed.
Delivery focus
Designed around the work that matters.
- 01
An HR knowledge web application with conversational questions and answers
- 02
An admin interface for concise knowledge entries and categories
- 03
One supporting attachment per knowledge entry
- 04
On-demand Sync to update searchable information
- 05
Context-aware query rewriting using the previous six conversation items
- 06
Content retrieval with Azure AI Search and answers composed by Azure OpenAI
- 07
AI-assisted decisions on whether to include an attachment from retrieved knowledge
- 08
AI workflow orchestration across context, query rewriting, retrieval, responses and attachment selection
Knowledge you can manage and update
Each knowledge entry focuses on one topic, allowing administrators to manage knowledge in smaller pieces and update individual subjects easily. One supporting file can be attached to each entry. After adding content, administrators can select Sync to update searchable information when needed.
From questions to answers with supporting information
The system retrieves content relevant to a question and uses it as context for Azure OpenAI to compose a conversational answer. This connects the knowledge base with AI through Retrieval-Augmented Generation (RAG). An AI decision step determines whether an attachment from the retrieved knowledge is needed to support the answer. Relevant images or files can then accompany the response, making explanations and related documents accessible from the same web page.
Conversational RAG and AI workflow orchestration
To support follow-up questions, the workflow retrieves the previous six conversation items and uses that history to reinterpret and rewrite the current question. This context-aware query rewriting helps carry the meaning of the conversation into knowledge retrieval. AI workflow orchestration connects context preparation, query rewriting, knowledge retrieval, response generation and attachment selection. A separate AI decision step assesses whether an attachment from the retrieved knowledge is needed to support the answer.
Build what matters
Have a similar challenge in mind?
Let's start with the outcome you want to create.
Start a Project