Connecting knowledge sources
We connect file shares, wikis, ticketing systems, and business applications through their interfaces. Documents are processed, split sensibly, and tagged with metadata such as version and department.
RAG knowledge assistantAn AI knowledge assistant
An AI knowledge assistant
that cites its sources.
Your staff ask questions in plain language and get answers drawn from your own documents. Every answer points to the passage it came from.
Book a free AI opportunity review
45 minutes, no strings attached. You leave with a written shortlist.
What it is about
RAG knowledge assistant
In most companies the knowledge exists, it is just hard to find. It sits in manuals, contracts, ticket histories, minutes, and the wiki. Finding it takes time and often depends on a handful of experienced colleagues.
A knowledge assistant based on retrieval-augmented generation (RAG) finds the relevant passages for each question and writes an answer from them. The language model does not add anything of its own; it refers to the source. If it finds none, it says so.
One point matters for company-wide use: existing permissions stay in place. If someone may not see a document in the source system, they will not get an answer based on it.
What we build
The building blocks
Search and cited answers
Search combines meaning with exact terms such as part numbers or clause references. Every answer includes links to the passages used, which open with a single click.
Permissions per user and department
Access rights from the source systems are carried over and checked on every request. One assistant can serve the whole company without confidential content leaking across teams.
Keeping content current
New and changed documents are picked up automatically, and deleted ones are removed. Outdated versions are flagged or excluded.
Evaluating answer quality
Using a set of typical questions with verified answers, we measure how often the assistant is right and cites correctly. User feedback feeds into ongoing improvement.
Approach
From the first conversation to operations
- 01
Free AI opportunity review
In 45 minutes we identify which knowledge sources and questions offer the most value. No obligation and no cost.
- 02
Pilot on one source
We start with one source and one user group, for example the service manual for your support team. Answer quality is measured on real questions.
- 03
Fixed-price roll-out
Further sources, permissions, interface, and integration are delivered as packages with a fixed price and acceptance criteria, such as a minimum share of correctly answered test questions.
- 04
Operations and maintenance
We monitor quality and usage, maintain the connectors, and adapt the assistant to new content. Or we hand it over to your IT team with full documentation.
By industry
Where this applies in practice
The industry overview lists concrete use cases. Detailed pages are available in German.
- Machinery and plant engineeringUse cases: Quotation and configuration support · Service and after-sales support · Preserving expert knowledge
- Logistics and freight forwardingUse cases: Supporting claims handling and dispatch
- Skilled trades
- Tax advisory and law firmsUse cases: Internal knowledge search
- Medical practices and outpatient centresUse cases: Internal quality management knowledge base
- Real estate and property managementUse cases: Owners meeting minutes and contract review
- Retail and e-commerceUse cases: Customer service assistant
- Industry and manufacturingUse cases: Preparing complaints and 8D reports · Knowledge assistant for work instructions
- Utilities and municipal energy providersUse cases: Internal knowledge assistant for regulations
- Insurance brokersUse cases: Preparing policy and terms comparisons
- Construction and architectureUse cases: Searching tender documents · Standards and project knowledge assistant
- Hotels and hospitalityUse cases: Multilingual guest communication · SOP and knowledge assistant for staff
- Agriculture and foodUse cases: Preparing IFS and QS audits
Related services
What often comes with it
- On-premise LLMsWe set up open-source language models so that your data stays inside your organisation or the EU. With permissions, audit logs, and controlled updates.
- AI agents & automationAn agent reads a request, turns it into a plan and carries it out across your systems. Before it changes anything, a person signs off.
- Document AIWe automate your incoming mail: documents are recognised, read, checked, and passed to your systems. Uncertain cases go to a person for approval.
FAQ
Questions about RAG knowledge assistant
How does ChatGPT work with your own company data?
The approach is called retrieval-augmented generation. Your documents are made searchable, the relevant passages are retrieved for each question, and the language model receives them as the basis for its answer. The model is not trained on your data.
Does an AI knowledge assistant make up answers?
The risk can be reduced substantially, but not to zero. We rely on source citations in every answer, an instruction not to answer without a matching source, and regular measurement against verified test questions. That keeps every statement checkable.
Which documents can an internal AI chatbot use?
PDFs, Office files, wiki pages, emails, ticket histories, and content from business applications with an interface. Scanned documents are made readable with text recognition first. Tables and drawings are tested separately in the pilot.
Can every employee see all content?
No. The assistant inherits permissions from the source systems and checks them on every request. Answers are only drawn from documents the person asking would be allowed to open directly.
How much does an AI knowledge assistant cost?
Mainly it depends on the number and type of sources, the permission model, and how it is hosted. A pilot on a single source is a contained piece of work. After the pilot we quote a fixed price per package for the roll-out.
How does the assistant stay up to date?
The connectors pick up new, changed, and deleted documents automatically. Every answer shows the version date of the cited source. Rules can be set for outdated versions so they are no longer used.
RAG knowledge assistant: let's find out in 45 minutes whether it pays off for you.
Book a free AI opportunity review
45 minutes, no strings attached. You leave with a written shortlist.