Summarising Documents with AI: The d.velop pilot in Action at Zentis

“The future of document management lies in intelligent process automation and the smart utilisation of existing organisational knowledge.” This is how Jan Föcking, Senior Software Development Engineer AI at d.velop, describes the significance of Artificial Intelligence (AI) in today’s business landscape. He also refers to the use case of food manufacturer Zentis, particularly in the summarisation and extraction of key information from complex contractual documents.

Summarising documents using Artificial Intelligence has become a widely adopted use case. Thanks to innovative technology in document management, it significantly boosts efficiency, reduces errors, and enhances knowledge management across the organisation.

How AI is Transforming Document Summarisation – Challenge, Solution & Potential

Whether at work, university or in everyday life – summarising documents and extracting key information from lengthy, complex texts has always been a time-consuming and tedious task. Often, it ends with entire paragraphs highlighted in neon colours, making it difficult to keep track of the essential facts, let alone process them correctly. Added to this is the challenge of handling sensitive information in a data protection-compliant manner and ensuring that unauthorised individuals do not gain access.

Digitalisation and innovative technologies such as Artificial Intelligence have become everyday heroes in this context. The processing and storage of digital documents in a document management system already resolve the issue of security through audit-proof and legally compliant archiving in accordance with GoBD. However, the manual effort of summarising remained – until the introduction of AI.

Advanced generative AI algorithms now support secure and efficient document processing. The technology analyses and understands the content of a document and, depending on the task (prompt), extracts the most important information. The level of detail in the summary can be defined. The AI either pulls relevant content directly from the original text or generates a shorter version in its own words – delivering a summary in just a few seconds.

Over half of respondents believe AI can make life easier

According to a study by Bitkom, 55% of respondents believe that Artificial Intelligence can simplify everyday life. However, the actual use of generative AI applications paints a different picture. Among private individuals, generative AI is already widespread – half of the population has tested it. In contrast, its use in businesses remains relatively limited. Only one in ten companies currently uses it, for example, to generate text, software code or images. The potential is therefore considerable.

The days when companies had to ask themselves whether generative AI could offer them any benefits are over. Today, it’s no longer about if – it’s about how, when and where.

Dr. Ralf Wintergerst
President of Bitkom

Summarising documents with AI: d.velop pilot – features & benefits

The innovative AI assistant from d.velop integrates seamlessly into existing document management systems. Via the chat function, the d.velop pilot can answer questions about one or multiple documents and is capable of generating summaries or identifying temporal relationships between documents.

In addition, intelligent assistants can be configured with ease to process organisation-specific domain knowledge and provide users with access to this information in natural language.

The d.velop pilot leverages state-of-the-art Artificial Intelligence technologies. Inputs in the form of natural language queries are interpreted using Large Language Models and translated into actions.

Functions at a Glance

• Intelligent collaboration with document management: Seamlessly integrated into the DMS, the AI assistant supports efficient management and organisation of digital documents, enabling direct interaction with them. It simplifies the search and categorisation of document information and enhances user experience.
• Intelligent context: The AI assistant recognises documents and their content, automatically generating contextual information for each document – such as practical short summaries.
• Intelligent levels of abstraction: A valuable feature is the ability of the AI to summarise content at different levels of abstraction. For example, the key points of a scientific paper can be presented in a way that is understandable even to non-experts.
• Intelligent chat: The chat function allows users to ask questions about a document or enter queries. Summaries, contract deadlines and other relevant information are provided in real time.
• Intelligent processes: In combination with the AI assistant, the AI-supported process automation of the d.velop process studio can efficiently handle routine tasks.

Benefits and reasons to use AI-powered document summarisation from d.velop:

• Efficiency and productivity: Fast access to knowledge saves time and costs. Replacing manual tasks and routine work frees up resources for value-adding activities.
• Reliability and control: Based on organisational knowledge, the evaluated information and summaries are dependable. All suggestions can be easily overridden by the user.
• Error reduction: The AI assistant’s precise and consistent summaries help minimise human error.
• Knowledge management: Data and information are not lost or buried in archives – they remain accessible, ensuring that knowledge is always available and information security is maintained.
• Availability: Thanks to seamless integration into the existing application landscape, the functionality is always readily available within the document management system.
• Compliance: Adherence to security standards is ensured.
• Practical relevance: Support is provided through employee self-service, product engineering bots, sales assistants for quote creation, and bots for service and support processes.
• Up-to-date: Continuous development ensures the assistant remains at the cutting edge of technology.

Use Case: AI in Action at Zentis – Summarising Documents, Automating Processes & Unlocking New Possibilities

d.velop blog: Is there a concrete use case for document summarisation with AI among our users? How have they benefited from using the d.velop pilot?

Jan Föcking: We currently see AI-based summarisation as particularly valuable wherever people need to read and assess long and/or complex documents in full. One such case is currently underway with the food manufacturer Zentis.

Even today, Zentis continues to demonstrate a strong sense of innovation and is using the d.velop pilot in a forward-thinking way to support knowledge management.

In the current project, we’re supporting highly skilled knowledge workers in the legal department to process and review contracts more efficiently. As soon as a contract is imported, the d.velop pilot summarises it and extracts key information from the document, significantly simplifying the subsequent handling of the contract for those responsible.

d.velop blog: How is the d.velop pilot being used at Zentis?

Jan Föcking: It’s quite an exciting case. Initially, the customer used the standard chat function and quickly realised that the d.velop pilot could summarise the key elements of a contract and extract specific information such as contracting parties, duration, deadlines, jurisdiction, confidentiality clauses, etc. However, the customer soon approached us with the request to go beyond chat-based interaction. While chat was a good starting point, they saw potential for even greater relief through automation.

Using the d.velop process studio, the customer was able to build their own form, allowing users to simply upload a contract. Behind the scenes, the form triggers specific modules provided by the d.velop pilot within the process studio. This way, the customer created their own tailored solution, with the d.velop pilot working in the background to extract the required contract information.

d.velop blog: Are there any new features or developments in the d.velop pilot?

Jan Föcking: Yes – we’ve recently introduced prompt suggestions, which are always available in the chat. Popular actions, such as document summarisation, are now just one click away.

We’ve also significantly improved summarisation across multiple documents. Additionally, the d.velop pilot can now identify temporal relationships across documents – for example, when analysing trends in quarterly or annual reports over several periods.

Soon, we’ll be offering a version of the d.velop pilot that uses models not hosted by hyperscalers (large cloud providers). This offering is especially relevant for regulated industries.

We’re currently working on implementing new source referencing. In future, the d.velop pilot will explicitly indicate which document a given statement is based on – and cite it accordingly.

d.velop blog: What trends and hypes are currently emerging in AI-powered document management? What future developments can we expect?

Jan Föcking:

• Agentic AI: Moving beyond assistants that support knowledge workers, we’re heading towards agents that not only provide access to knowledge but also derive actions and autonomously complete entire tasks.
• Data quality as a differentiator: “Garbage in – garbage out” still applies. Without a sufficiently high-quality data foundation, even the best AI models can’t deliver good results.
• Smaller, specialised AI models: For specific industries and use cases, we’ll see significantly smaller, specialised models – some of which can be operated locally.

Conclusion – AI Agents Are the Future of Document Summarisation

How, when and where AI will become part of everyday life is something each organisation must decide for itself. What’s important to take away is that AI, as a supportive tool, is revolutionising both the way businesses handle information and the automation of routine tasks, knowledge management, and data extraction. The example of confectionery manufacturer Zentis clearly shows that AI can summarise documents – saving time while significantly improving efficiency and accuracy.

The integration of AI into document management, as demonstrated by the d.velop pilot, clearly illustrates how organisations can benefit from this technology.