Automation and AI solutions
RAG for business course
Assistants that answer from your company's documentation, citing the source.
Most of a company’s knowledge lives in documents: procedures, contracts, manuals, resolved tickets. With RAG (retrieval-augmented generation) a language model can answer using that documentation, citing the source, without training any model.
A RAG demo is easy; making it work well with real documentation is not. This course focuses on what makes the difference: preparing documents properly, combining semantic and keyword search, respecting each user’s permissions and, above all, evaluating the answers so you improve with data, not by eye.
More context in the article how an AI assistant works with your company’s documentation.
What you will learn
- Understand the RAG architecture and when it is the right solution
- Prepare and chunk documents so they are retrieved well
- Choose the embedding model, vector database and search strategy
- Respect access permissions to information
- Evaluate answer quality and reduce made-up answers
Syllabus
8 modules, from the fundamentals to the final project.
-
Why RAG
- The limits of a language model without context
- RAG versus fine-tuning and versus long context
- Business use cases
-
Preparing the knowledge
- Extracting text from PDFs, Office files, wikis and tickets
- Chunking strategies
- Metadata, dates and versions
-
Embeddings and search
- What embeddings are
- Vector databases and pgvector
- Hybrid search (semantic and keyword) and re-ranking
-
Generating the answer
- Instructions to answer only from the context
- Source citations
- Answering "I don't know"
-
Permissions and security
- Filtering by user and group
- Sensitive data
- Instruction injection via documents
-
Evaluation
- Set of reference questions and answers
- Retrieval and faithfulness metrics
- Automatic evaluation with models
-
Advanced patterns
- Query rewriting and multi-step search
- RAG with agents
- GraphRAG and structured data
-
Production
- Keeping indexes up to date
- Cost, latency and caching
- Quality monitoring
Tools you will practise with
- Python
- LangChain or LlamaIndex
- pgvector, Qdrant or Azure AI Search
- Embedding models
- Ragas
Hands-on project
You build a RAG assistant over a set of real (or sample) documents, with source citations and permission control, and measure its quality with an evaluation set before and after improving it.
Requirements
- Basic Python programming.
- Basic knowledge of APIs.
Other courses
-
AI agents for business course
Agents that act on your systems, with clear limits and human oversight.
- Duration: 12–16 hours
- Level: Advanced
-
AI automation with low-code tools course
Automate complete processes by combining AI and visual tools, without coding.
- Duration: 12 hours
- Level: Beginner
-
Copilot Studio and Power Automate course for citizen developers
Build AI agents and automations in Microsoft 365 without coding, and within the rules IT needs.
- Duration: 12 hours
- Level: Intermediate
Where it is applied
-
Chatbots and internal assistants
We build chatbots and internal assistants connected to your company's information to answer questions, reduce repetitive queries and help the team find what it needs faster.
-
AI internal assistant for company documentation
We build internal assistants that let your team search documents, manuals, procedures and knowledge bases in plain language.
Related articles
-
Use cases
AI internal assistant for company documentation (RAG)
How an AI internal assistant that answers from your company's documentation (RAG) works, when it is worth it, the risks and how to run a pilot.
7 min read
Want this course for your team?
We adapt the content, length and format to your company.