Industry

Expert Knowledge Capture Agent

What your colleague knows about the line retires with him on Friday. Nobody ever wrote it down.

Turns knowledge capture on its head: the agent asks, the experienced colleague just talks.

Open agent
Public administration

Four processes that take up the most time today

Developed in Germany · Hosted in the EU

Developed in Germany · Hosted in the EU or on-premises

Delivered together with

STACKIT Sovereign hosting in Germany, single-tenant from the first user.
Rödl Legal, tax and compliance expertise for legally sound deployment.
4P Consulting Advisory and implementation for the public sector, starting with the AI readiness assessment.

Twelve weeks of handover is not enough when the writing is supposed to happen alongside the shift.

The handover is planned as an overlap period: the experienced line operator is supposed to document what he knows. In everyday work, a breakdown gets in the way, and what ends up recorded is what the maintenance schedule says anyway. The special knowledge, the fault pattern with its three exceptions, stays in his head and leaves with him.

38years of serviceThe line operator has known the line since it was commissioned.
12weeks of handoverThat is how long he overlaps with his successor.
6fault patterns with no entryThey appear in no maintenance schedule, because they occur so rarely.
9months of onboardingThat is how long the successor needs before he can solve the same cases on his own.

On the last working day, the six fault patterns sit as searchable entries in the knowledge base instead of in the colleague's head: up to 60% less knowledge loss.

The reduction of up to 60% in knowledge loss is this agent's headline figure. The six fault patterns without an entry and the nine months of onboarding sketch a typical handover; nobody counted them.

Three steps through the process.

The experienced colleague decides himself whether each entry is released; the agent asks the questions and puts what he says into shape.

Narrow down the knowledge topics in conversation

The agent goes through the plants and fault patterns the colleague was responsible for. It flags the topics for which there is no entry.

Topic list for Line 4, 23 topics, 6 with no entry in the maintenance schedule

Draw out the experience in an interview

For every open topic, the agent asks follow-up questions until trigger, warning signs and procedure are named. The colleague talks; notes are taken as he speaks.

47-minute interview, 6 fault patterns captured with trigger and procedure

File the entries as drafts

The agent files every topic as a draft with the plant reference and search terms. The colleague reads it through and releases what is correct.

6 drafts in the knowledge base, each with 1 plant number and 4 search terms

What the agent does not do.

It does not release any entry itself.

It presents the draft with the plant reference. Whether it goes into the knowledge base as it stands is for the colleague who described it to sign off.

It does not judge whether a procedure described is technically correct.

It records what the colleague describes and marks the passage as experience-based knowledge. The technical assessment is for the head of maintenance.

It only asks about the plants that come up in the conversation.

An auxiliary unit the colleague never mentions never comes up in any interview. Knowledge management decides which topics belong on the list.

Connected systems

  • Knowledge base
  • maintenance schedules in the maintenance system
  • onboarding plan in the HR system

Operations

  • Deployment modelEU cloud or private cloud
  • Data processingGDPR-compliant, no model training on your data
  • Responsible teamsService, HR, knowledge management

The questions that come up before the first meeting.

How quickly is the Expert Knowledge Capture Agent ready to use?

A pilot typically runs on your own data within 6 to 8 weeks. The agent connects to your existing systems rather than running as a separate, parallel environment.

Where is the data processed?

EU cloud or private cloud. Processing is GDPR-compliant and models are never trained on your data. Details in the Trust Center.

Where does the figure “up to 60% less knowledge loss” come from?

It is based on experience from comparable projects and process benchmarks, not a commitment. We establish the realistic effect for your organisation in the assessment before the pilot.

Related agents.

Intake review, case 2026/04711

Skip to content

Page footer