AI chatbot for companies with source references
Internal knowledge search and chatbots built around your content: approved answers or RAG with a language model. T-NEX scopes access, data and implementation.
Learn moreArchitecture and implementation
Make your company's knowledge available where it is needed.
Documents, answers and practical experience become a usable knowledge base for your team. We combine search, editorial maintenance and suitable AI functions around your content. Sources and responsibilities remain linked to the knowledge.
Cobalt · Knowledge assistant
Find approved knowledge with its source.
Find a relevant answer and open its source. Maintained questions, answers and synonyms connect mobile search with knowledge administration.
Maintain knowledge entries
Choose a question
The mobile start page offers suggested questions about leave and onboarding alongside recent conversations. Selecting a question opens the knowledge assistant.
A specific question leads into the chat.
Enlarge viewRead the answer and source
A maintained answer explains leave requests and carrying leave forward. The visible source reference names Leave Policy v3.2, section 4.
The source reference stays visible with the answer.
Enlarge viewMaintain knowledge entries
The administration view places answers, categories, source documents and search terms side by side. Entries can be added and edited directly.
Answer text and its source are maintained together.
Enlarge view In daily work
Knowledge management starts with users’ questions
A knowledge system should answer practical work questions: which instruction applies, why a decision was made and where the original information can be found. T-NEX connects source selection and maintenance with an appropriate search or answer interface.
We begin with a defined collection of documents, domain terms and typical questions. Provenance, currency and access permissions are made explicit. Missing or contradictory sources are addressed before an AI connection is added.
Overview
Choose an appropriate mix of document search, RAG and a maintained wiki
Document search returns source locations. RAG adds a composed answer. A maintained wiki organises and consolidates information into linked knowledge pages. A knowledge graph explicitly records relationships between entities. These approaches address different parts of the knowledge task.
We compare suitable approaches against your questions. A stable core of frequently needed rules can be curated, while less common detailed questions use original documents. This combination is tested for usefulness rather than assumed to be universally more accurate.
| Approach | Useful role in the workflow | Maintenance need |
|---|---|---|
| Document search | Find original passages | Document collection, versions and access |
| RAG | Combine retrieved content into an answer | Review sources, retrieval and answer quality |
| Maintained wiki | Organise and link frequently needed knowledge | Update pages and resolve contradictions |
| Knowledge graph | Represent relationships between entities explicitly | Maintain entities and relationships |
Overview
Keep original sources connected to curated knowledge
A knowledge page should lead back to the documents from which its statements were derived. Terms, responsibilities and relationships are described so that an individual passage remains understandable. This supports both search and domain review of an answer.
Originals remain important: tables, attachments and differing document versions often contain details omitted from a summary. Review must therefore include a path from the result back to the underlying source.
Overview
Carry access, updates and ownership through the system
Content owners are assigned to sources and knowledge areas. A change to a work instruction can trigger review of linked knowledge pages and test questions. Access groups determine which information each audience is permitted to use.
Maintenance is established as an ongoing process, including a traceable change history and a way to report missing or unclear answers. A well-designed search interface alone does not keep the collection current.
Overview
Evaluate answer quality against a fixed question set
The comparison uses typical questions, uncommon detailed questions and cases with contradictory or missing sources. Domain specialists assess the content, supporting sources and limits of an answer. Testing also checks that different users receive only information intended for them.
The same question set can be used to assess later changes, revealing whether a new import or search configuration degrades previously useful answers. T-NEX itself uses a linked knowledge base built from original documents and brings that working method into implementation.
FAQ
Questions about the solution
Do we first need a complete company wiki?
No. A defined knowledge area with clear users and owners is sufficient to start. The structure should grow from real questions and reviewed content.
Is a wiki better than RAG?
It depends on the questions, sources and maintenance. A wiki can consolidate frequent knowledge; RAG can access larger document collections. Both can overlook information or use outdated material. The decision is made by comparing them for the actual use case.
What does ongoing maintenance include?
Adding new and changed sources, resolving contradictions, updating linked knowledge pages and rerunning important test questions. Each relevant knowledge area needs a domain owner.
Can confidential areas be separated?
The intended access groups are defined during architecture work and checked during retrieval. Separation must apply to search results, answer content and linked original sources.
Related options
You may also be interested in these.
Document AI
Document processing as a custom project: define required fields, test extraction against examples and connect the next step in your workflow.
Explore serviceAn AI pilot with defined evaluation criteria
Evaluate one AI use case with T-NEX: limited scope, suitable data and agreed criteria for expansion or stopping.
Explore serviceWorks councils and AI
Prepare workplace AI in Germany: early information, co-determination, employee data, pilot boundaries and a system-specific works agreement.
First version: Updated:
Read articleWhich task would you like to solve next?
Bring a concrete task. Together, we will define what the application needs to do.
Discuss a knowledge task