How external data makes AI more reliable - Retrieval augmented generation explained

Artificial intelligence (AI) is changing the world of work - but many companies are faced with the challenge of incomplete or unreliable answers from language models. Retrieval Augmented Generation (RAG) offers a solution: the method combines AI systems with external, verifiable data sources to create comprehensible, up-to-date and fact-based results.

Translated with DeepL

SMEs in Switzerland in particular benefit from this technology. RAG improves customer service, strengthens knowledge management, accelerates internal processes and supports fault diagnosis in production.

In the new article, Prof Dr Mascha Kurpicz-Briki from Bern University of Applied Sciences explains how companies can link their own data with AI in a targeted manner - and why RAG is the key to trustworthy artificial intelligence.

Read full article (in German)

Context:

This article in KMU-Magazin is part of a series that has emerged as a follow-up activity to the publication Orientation AI: Challenges and Opportunities for Swiss SMEs. This SATW publication supports small and medium-sized enterprises in recognising the potential of artificial intelligence and planning concrete next steps.

The content was created in close collaboration with stakeholders from the SAIROP (Swiss AI Research Overview Platform) network. SAIROP promotes the exchange between science, business and society, makes Swiss AI expertise visible and provides orientation in the dynamic AI ecosystem.

The SATW is committed to Digital Trust - become part of this development and contact us:

 Manuel Kugler

Manuel Kugler

Data & AI Programme Manager / Advanced Manufacturing

Questions and answers covered in the article:

RAG is a technology that combines artificial intelligence (AI) with the company's own knowledge. Language models such as ChatGPT are linked to internal documents, databases or manuals in order to provide more precise and comprehensible answers.

Many SMEs have valuable knowledge in the form of PDFs, instructions or emails - but this is difficult to access. With RAG, this knowledge can be automatically searched and processed in natural language. This creates a digital assistant that relieves employees and answers customer enquiries more quickly.

Traditional language models sometimes "invent" answers because they are based on statistical patterns. RAG, on the other hand, is based on real data sources - this reduces errors, increases transparency and strengthens confidence in the results.

  • Customer service: Automated, fact-based answers to frequently asked questions
  • Internal knowledge search: employees find information faster
  • Technical support: assistance with maintenance, fault diagnosis or product explanations
  • Time and cost savings thanks to automated information processes
  • Higher service quality thanks to precise answers
  • Better basis for decision-making thanks to up-to-date, internal data
  • Strengthening competitiveness through intelligent knowledge utilisation

The introduction of RAG requires a clean database and responsible integration. Data protection, data quality and the ongoing maintenance of knowledge sources remain key success factors.