Finding information is only one part of modern knowledge work. It is equally important to assess relevant documents, combine information from multiple sources, and understand what a statement is based on. INTERGATOR 6.8 therefore advances both search and AI-supported work with enterprise knowledge.

The main enhancements include the new Agentic Chat, improved hybrid search, and extended AI tasks. Further improvements affect source references, usability, accessibility, and the operation and maintenance of the platform.

Search, chat and AI tasks serve different purposes. Search is designed to find relevant documents and information and to narrow down result sets. Agentic Chat is intended for specific questions and dialog-based research. AI tasks, in turn, support recurring analyses of individual documents or search results.

Agentic Chat organizes research more flexibly

One of the most significant changes in INTERGATOR 6.8 concerns the chat. Previously, processing followed a more predefined sequence: a search returned documents that were then used to generate an answer. The new Agentic Chat works more flexibly. The language model first analyzes the user input and decides how the request should be handled depending on the context. It can, for example, answer directly, use INTERGATOR search, or ask a clarifying question if important information is missing.

Depending on the model being used, the chat can also perform several searches in sequence and combine the results into one answer. This allows more complex questions to be researched step by step without requiring users to formulate every individual search query themselves. The term “Agentic” does not mean that INTERGATOR acts as an autonomous AI agent capable of carrying out arbitrary actions. In version 6.8, the primary tool available to the chat is INTERGATOR search. Professional review of results and business decisions remain the responsibility of the user.

Search and chat serve different purposes

With INTERGATOR 6.8, the distinction between structured search and dialog-based information access becomes increasingly important.

Search is the appropriate starting point when users want to find documents or information on a topic, build a result set, and then narrow it down using filters or facets.

A search might begin with a precise phrase such as:

“hydrogen propulsion ocean shipping”

If the information need is broader, users can also provide a more detailed description, for example:

“technical challenges of using hydrogen propulsion in ocean shipping”

Agentic Chat, by contrast, is intended for specific questions such as:

“What challenges are associated with the use of hydrogen propulsion in ocean shipping?”

The difference lies primarily in the expected result. Search provides a structured list of results that users can review, filter and process themselves. Chat attempts to combine information from suitable sources and provide a formulated answer. Users can then continue the research with follow-up questions. This distinction helps users select the appropriate tool for their specific information need.

Hybrid search combines terms and meaning

Search itself has also been further developed in INTERGATOR 6.8. In the AI editions, hybrid search combines lexical and semantic search methods. Lexical search is particularly useful when users are looking for specific terms, names, numbers or known phrases. Semantic search focuses more strongly on the meaning of an input and can therefore also identify documents that describe a topic using different terminology.

INTERGATOR combines both approaches. This means users can work with individual search terms as well as longer natural-language descriptions of an information need without having to switch between different search modes. This is particularly useful when the exact wording used in a document is unknown or when relevant content is available in different languages.

In INTERGATOR AI+, relevant passages within a search result can also be identified. A language model helps determine which terms and text passages are particularly relevant to the respective search query. The displayed text itself continues to originate from the indexed documents.

Source references make chat answers more transparent

Generated answers are especially useful when users can understand which information they are based on. INTERGATOR 6.8 can therefore display inline references to documents used within a chat answer. This allows users to move directly from a statement to the underlying source. In addition, documents used for an answer can be displayed in a source list. From there, users can open either the document preview or the original document.

However, not every answer necessarily has to be based on enterprise documents. Depending on the request and configuration, Agentic Chat can also use the general knowledge of the underlying language model. Source references therefore also help users understand which statements are based on retrieved enterprise information. Important information should nevertheless be verified against the original documents. Source references improve transparency but do not replace professional review.

Chat scopes remain available for focused research

With the new chat logic, users no longer necessarily have to select a specific knowledge area before starting every conversation. Chat scopes nevertheless remain available. They can be used to restrict a conversation to a defined set of documents, such as selected documents, favorites or lists.

This is useful when a question should deliberately be answered only on the basis of a particular collection of documents. At the same time, starting a new chat becomes easier when such a restriction is not required. The available chat scopes and document areas depend on the respective configuration and user permissions.

Answers appear while they are being generated

Chat answers in INTERGATOR 6.8 are displayed while they are being generated. Users therefore no longer have to wait until a complete answer has been generated before they can start reading. Particularly for more complex research tasks or more capable language models, this improves the perceived response time.

The time required to produce the complete answer still depends, among other things, on the model being used, the number of searches required, and the complexity of the request.

AI tasks become easier to reuse

In addition to search and chat, AI tasks have also been enhanced. They allow defined instructions to be applied to search results or to an open document. Frequently used tasks can now be pinned. This makes them directly available again at the respective point of use. Individually formulated questions or instructions can also be saved together with their wording for repeated use.

If a larger number of tasks is available, the task list can additionally be filtered by title or description. Typical use cases include summarizing a document, identifying key topics or facts, and carrying out project-specific review and analysis tasks. Unlike chat, AI tasks are not dialog-based. They apply a defined instruction to a specific document context and are therefore particularly suitable for recurring processing steps.

More consistent usability and document access

Several areas of the user interface have also been made more consistent. For documents displayed in favorites and lists, the title now links directly to the original document, while the preview image opens the document preview in INTERGATOR. This makes it easier to decide whether a document should be opened directly or first reviewed within the platform.

Further improvements have also been made to the accessibility of the user interface. Which elements and functions are visible continues to depend on the edition, configuration and customer project.

Improvements behind the scenes

Some of the changes in INTERGATOR 6.8 are not directly visible to end users but improve the operation and maintainability of the platform. These include updated components and base images, improved logging and error messages, and more robust operational processes. The processing of embeddings has also been simplified. Document and query vectors are generated in a way that ensures the underlying configurations remain compatible. This reduces configuration effort and technically safeguards an important requirement for reliable semantic and hybrid search.

Security and maintenance processes have also been enhanced, and individual connectors have been modernized. Which connectors and operating options can be used in a specific installation depends on the respective system environment and project configuration.

What INTERGATOR 6.8 changes for knowledge work

INTERGATOR 6.8 expands the possibilities for AI-supported research without replacing traditional search. Hybrid search supports the targeted discovery and filtering of information. Agentic Chat adds a dialog-based approach in which users can ask specific questions, trigger searches and combine information from multiple sources. AI tasks, in turn, support recurring analyses at document and result level.

This creates different ways of working with the same knowledge base: searching and filtering, researching in dialog, or processing documents according to recurring criteria.

Which approach is most appropriate depends on the specific information need. The key principle remains that enterprise information comes from connected and approved data sources, existing permissions are respected, and important results can be verified against the underlying documents.