Thirty years of experience with legacy filing systems, DMS and intranets · TÜV-certified AI competence

AI Knowledge Management: Securing, finding and sharing knowledge

When the same questions always end up with the same people, your company doesn't have a knowledge problem. It has an access problem. We'll show you which knowledge can be unlocked from documents, systems, and people's minds – and what needs to be clarified first.

Three people in a meeting at a laptop

When knowledge leaves the company with a person.



A customer asks about a custom solution from 2018. What was delivered is recorded in the ERP system. Why it was built that way and what shouldn't be changed is only known by an experienced colleague – and they are retiring next year.

At the same time, other employees repeatedly answer the same questions, search for old emails, or compare multiple versions of a document. This takes up a bit of time every day and is therefore rarely recognised as a real problem.

During holidays, illness, growth, or succession, this dependency suddenly becomes visible.


Another wiki rarely solves this. Knowledge only remains usable if it is created during everyday work, is easy to find, and someone remains responsible for keeping it up to date.

What AI knowledge management can do for you

It shortens the path to reliable information.

An AI-supported knowledge system provides an understandable answer to a specific question – together with the documents and passages on which it is based. It does not replace a professional decision.

In doing so, the knowledge can remain in its existing location: in the DMS, intranet, ticket system, or in shared repositories. An additional layer opens up the sources and takes existing permissions into account.

Workshop at a wall covered with colourful sticky notes

Request a free initial consultation

Tell us a question that always ends up with the same person in your business. We'll give you an initial assessment of where the answer currently lies and whether a structured approach is worth pursuing.

Four types of knowledge – four distinct tasks

Documented and up to date
  • Typical source: manuals, work instructions, product data

  • What is needed: share and connect the source

Documented, but outdated
  • Typical source: Old folders, multiple versions

  • What is needed: Clarify validity and responsible parties

Scattered
  • Typical sources: emails, minutes, tickets, quotes

  • What is needed: select sources and define permissions

Only in people's minds
  • Typical source: experiential knowledge of individual employees

  • What is needed: organise structured knowledge transfer

Documented and up to date
  • Typical source: manuals, work instructions, product data

  • What is needed: share and connect the source

Scattered
  • Typical sources: emails, minutes, tickets, quotes

  • What is needed: select sources and define permissions

Documented, but outdated
  • Typical source: Old folders, multiple versions

  • What is needed: Clarify validity and responsible parties

Only in people's minds
  • Typical source: experiential knowledge of individual employees

  • What is needed: organise structured knowledge transfer

The critical issue is not a lack of knowledge, but contradictory knowledge. A convincingly phrased answer from an outdated document is more dangerous than no answer at all. That is why we start with sources and responsibilities – not with a chat window.

How to tell when action is needed

When it comes to certain questions, everyone mentions the same colleague.
When it comes to certain questions, everyone mentions the same colleague.
Holiday or illness of individual persons delays decisions.
Holiday or illness of individual persons delays decisions.
New employees have regular queries for months.
New employees have regular queries for months.
Solutions for recurring cases are repeatedly developed from scratch.
Solutions for recurring cases are repeatedly developed from scratch.
Multiple documents exist on a topic, but nobody knows which is the valid version.
Multiple documents exist on a topic, but nobody knows which is the valid version.
There is no structured knowledge transfer in the event of succession or retirement.
There is no structured knowledge transfer in the event of succession or retirement.

The more points apply, the more sense it makes to create a knowledge map before discussing software.

Consultation as a first step

Knowledge map

We collect recurring questions and assign them to areas of expertise. For each area, we document where the answer can be found, who is responsible for it, and how critical an outage would be.

Source evaluation

We review relevant repositories and systems: What can be used directly? Where are text recognition, access rights, or unambiguous versions missing? Which sources should be deliberately excluded?

Prioritisation and Business Case

We assess the benefits, preparation effort, technical feasibility and ongoing maintenance. This reveals which knowledge domain offers the most worthwhile starting point.

Roadmap

You will receive a concrete sequence of steps outlining the prerequisites, effort and costs for potential implementation. The roadmap is yours and can also be implemented internally or with another partner.

This is what implementation can look like

Step 1

Select a subject area

We start with a clearly defined topic, often where many questions arise or where a key expert will soon be leaving.

Step 2

Prepare and integrate sources

Valid documents, access rights, and rules on keeping content up to date are defined. If necessary, scanned files are processed using text recognition (OCR), or knowledge gaps are documented.

Step 3

Practise with real questions

A small group of users tests the responses in everyday practice. We measure the quality of results, usage levels and time saved based on predefined criteria.

Step 4

Anchoring care and responsibility

Every knowledge area needs an assigned owner, a review date, and a reporting channel for incorrect or outdated answers. Otherwise, you just end up with a technically more modern wiki that will eventually become outdated as well.

Three people in a meeting at a laptop

Capture knowledge from people's minds

Experiential knowledge cannot be extracted automatically.

When a key knowledge holder leaves, we conduct structured discussions about decisions, special cases, and typical mistakes. What matters is not only what someone does, but how they recognise what needs to be done in each specific case.

These discussions result in an organised version that the knowledge holder reviews and supplements themselves. This process should be communicated as a recognition of their experience – not as an attempt to make a person replaceable.

Which sources are typically suitable

Office files and text-based PDFs: Usually ready to use, provided their validity has been confirmed
Office files and text-based PDFs: Usually ready to use, provided their validity has been confirmed
Scans without text recognition: Usable after OCR
Scans without text recognition: Usable after OCR
Intranet and ticket system: Easy to use once areas and permissions are defined
Intranet and ticket system: Easy to use once areas and permissions are defined
Mailboxes and chats: Technically accessible, but organisationally and legally sensitive
Mailboxes and chats: Technically accessible, but organisationally and legally sensitive
Tables: Content usable; complex formula logic may need to be explained
Tables: Content usable; complex formula logic may need to be explained
Drawings, CAD and photos: often only useful in conjunction with descriptive text
Drawings, CAD and photos: often only useful in conjunction with descriptive text
Meeting recordings: Usable after transcription and with clear rules
Meeting recordings: Usable after transcription and with clear rules

The specific legal assessment, particularly regarding mailboxes, employee data or recordings, is a matter for data protection, legal counsel and, where applicable, the works council.

What changes for your business

New employees become independent faster

Experienced colleagues answer fewer repetitive questions

Holidays and sick leave are less likely to bring knowledge sharing to a standstill

Previous solutions are reused rather than reinvented

Knowledge transfer before retirement becomes plannable

During succession, financing, or a sale, dependencies become visible and can be reduced

What AI knowledge management does not achieve

An assistant does not create knowledge that does not exist anywhere. It does not decide which conflicting document is technically correct. It does not replace induction and training, nor does it replace a qualified professional. And it only remains reliable if sources, permissions, and relevance are properly maintained.

That is why we do not promise a ready-made knowledge base in just a few weeks without having seen your existing filing system. We start with an honest assessment.

The Moltomedia Difference

Experience with legacy systems

We know the reality of legacy IT landscapes: outdated folders, unclear permissions, scanned PDFs, in-house developments, and processes that only a single person fully understands. When it comes to knowledge management, this experience matters more than a flashy interface.

Strategy and development together

With us, IT strategy and development work hand in hand. This ensures that benefits, sources, permissions, integration, operations, and maintenance are planned as a unified initiative. We are not tied to any specific AI model and, wherever sensible, build the knowledge base so that the model remains interchangeable in the future.

TÜV certificate “AI Strategy & Application Expert” for Stella Pazzi

AI Strategy & Application Expert

2026

Logo of the German Federal Ministry of Research, Technology and Space

Funded by BMFTR

2025–2027

brand eins award: Best IT Service Providers 2026

brand Award 2026

2026

German Web Awards of the Year 2024 award

German Web Awards 2024

2024

“Innovative through Research 2022/2023” seal of Moltomedia GmbH & Co. KG

Innovative through research

22/23

Consultation between two people at a table

Frequently asked questions

Do we need a wiki first or perfect documentation?
How much does the consultation cost?
Can it be set who can see which content?
Is our content used for model training?
Who will maintain the knowledge base later?
How do we measure success?
Does this work across multiple languages and locations?

Start with a question, not the whole filing system

Send us a question that regularly ends up with the same person in your company. In a free initial consultation, we will clarify which sources are relevant for this and whether a knowledge analysis is worthwhile.

Get the clarity you need to move your business forward now.

Start today with your free IT assessment and get a clear roadmap for your digital future in just a few days.

© 2026 Moltomedia. All rights reserved.

Get the clarity you need to move your business forward now.

Start today with your free IT assessment and get a clear roadmap for your digital future in just a few days.

© 2026 Moltomedia. All rights reserved.

Get the clarity you need to move your business forward now.

Start today with your free IT assessment and get a clear roadmap for your digital future in just a few days.

© 2026 Moltomedia. All rights reserved.