The DiFA 2.0 project is a research cooperation between the VLBA and Volkswagen AG. The research co-operation is divided into two projects. Each project pursues different objectives.

Sub-project 1 - DocNexus

The "DocNexus" sub-project analyses the existing conditions in internal audit departments with regard to the management and use of unstructured and semi-structured data. This data includes process standards, audit reports and other audit-relevant documentation. The current situation shows that the existing methods for searching and compiling the necessary documents for audit preparations require a considerable amount of time and labour. This is mainly due to the existing limitations of keyword searches and query-based information retrieval. DocNexus uses various AI technologies as part of its technological implementation: generative AI, multimodal models, large language models (LLMs) and retrieval augmented generation (RAG). The research focus is on the methodical development of solutions for processing and extracting information from large unstructured data sets. The objective is to improve daily audit processes through the use of systematic information processing and structured retrieval systems.

Sub-project 2 - Use of AI in audit planning

This sub-project will investigate the use of artificial intelligence (AI) in audit planning in internal auditing. The aim is to design a kind of new audit universe that makes risk-oriented audit planning more efficient and targeted. AI will be used to automatically identify and prioritise relevant audit areas, analysing various data sources to identify potential risks and weaknesses. The new Audit Universe will enable Internal Audit to deploy resources in a more targeted manner and promote data-driven decision-making to ensure even more effective and precise audit work.

Volkswagen AG (Wolfsburg)

http://projekt-difa-v2.vlba.net/