DMS GPT-Agents
DocAnalyst
Automates the analysis of documents and enables the rapid extraction of relevant information, e.g. for the evaluation of application figures.
DocSeeker
Facilitates the targeted search for documents and their contents within databases.
DocMatcher
Links related documents, such as invoices with the corresponding delivery bills, in order to recognize correlations more quickly.
DocClassifier
Automatically recognizes and classifies documents, including indexing and pre-assignment, to optimize document recognition.
DocConfigurator
Enables easy configuration of workflows or process steps using intuitive voice commands.
DocLinker
Combines thematically related documents, e.g. all of a supplier’s documents, in one central location.
DocSyncer
Ensures synchronization between different systems, for example for automatic data transfer between financial accounting and ERP systems.
DocCreator
Automatically creates new documents, e.g. invitations or reports.
DocSupporter
Supports IT employees with AI-based support, especially for telephone inquiries and system support.
February 2025
On February 12, the second project meeting of Starke+Reichert GmbH & Co. KG and the Department of Information Systems at the University of Kassel as part of the Distr@al project GenKITs.
The project meeting not only served to present the results to date, but also to strategize the next steps.
As part of the completed work package 1, in addition to a collaborative process for identifying AI agents based on job roles, a large number of specific agents were also designed and prioritized as part of the workshop:
GenAI Pattern Canvas
September 2024
The patterns were developed in group work during the kick-off event. The team members worked in two groups to develop ideas for a document management solution supported by generative AI. This resulted in exemplary prompts as well as initial considerations regarding the necessary prerequisites and technical requirements. In addition, possible process changes were identified, the benefits of AI evaluated and the scalability of the solution analyzed. It was also determined which tasks would be performed by the AI and which by humans and to what extent the process would be positioned between human control and AI automation..
The use of patterns promotes an in-depth examination of the use case that goes beyond the mere collection of ideas. It also makes it easier for people who were not involved in the group work to understand the background, requirements and possible changes.
During development, general questions about the objectives of the process were clarified: What is being developed, for whom and for what purpose? It was determined in which area of the company the solution would be used and which processes or tasks could be improved as a result. After completion, the patterns were presented and prioritized. The order of prioritization is as follows:
- Intelligent search
- Link & Summarize
- Preselection
- Configuration automation
- Classification
Use Case Details
Intelligent document search
The aim is to simplify the search for documents using generative AI and make it more efficient. Currently, searching for and extracting content is often cumbersome and time-consuming. The AI should “understand” the context and content of documents and suggest suitable forms of presentation (e.g. tables or graphical overviews). This eliminates the need to manually sift through documents – a short query to the AI is enough to obtain a structured result.
Business Value: Time savings, increased convenience, reduced susceptibility to errors.
Link & Summarize
This use case aims to create a better overview of documents in the healthcare sector. Currently, manual summarization often leads to cognitive overload and relevant information is difficult to find. AI should provide support through structured diagrams or concise summaries and thus facilitate new diagnostic assessments.
Business Value: Time savings for doctors, improved diagnostic quality.
Preselection
AI is intended to optimize the initial screening process for application documents. Currently, screening is time-consuming and subject to inconsistent evaluation criteria. The AI filters applications based on predefined criteria, summarizes relevant content and can display this as a score, for example.
Business Value: Faster personnel selection, better comparability of applicants, general scalability.
Configuration automation
In this use case, the AI is intended to make manual and repetitive configuration tasks easier. Currently, special processes often have to be carried out manually outside the system, such as creating new forms. The AI translates natural language into automatic configurations, eliminating the need for manual clicks and providing intelligent suggestions instead.
Business Value: Time savings, reduced complexity for users.
Classification
The manual distribution and archiving of emails and documents currently requires a lot of work. AI automatically classifies content, recognizes the context and supports more efficient distribution and archiving.
Business Value: Reduced manual effort, improved data quality.
