What stays fixed
- which knowledge supports the statement;
- status and validity;
- conditions and exceptions;
- who is authorised to judge.
How DyReP works
A work request does not go straight to AI.
DyReP is the working layer between your organisation and AI.
DyReP first brings together the relevant context: who is asking, which knowledge applies, which way of working governs the task, what may be carried out and where a person must decide.
From work request to result in eight steps
- 01
Work request
Purpose, recipient and the result wanted.
- 02
Access
Who is asking, in what role, and what may that person access?
- 03
Knowledge
DyReP uses the approved knowledge available for this work request.
- 04
Way of working
Steps, criteria, tools and stopping points are fixed.
- 05
Execution
AI and tools do only what the way of working and authority allow.
- 06
Human judgement
Where judgement is needed, DyReP waits for the authorised person.
- 07
Form
The same foundation takes the form the recipient needs.
- 08
Result Passport
Where the application calls for it, the Result Passport records which work request, knowledge, way of working and decisions the result rested on.
The components involved, and the evidence available for each, are documented in one place.
See system and evidenceWork where you already work
Work in ChatGPT, Claude or your own environment.
The working environment can change. Knowledge, way of working and authority stay managed in one place, in DyReP.
Sources can also be submitted from that environment: a file goes through a submission window to the knowledge base, where it is checked and registered first. Being allowed to submit does not mean being allowed to review or include; inclusion in the knowledge base requires a separate human decision.
Which environments are demonstrably connected at the moment is shown in the evidence register.
See current connectionsDynamic re-Presentation
What can change and what must stay fixed.
What can move
- level of detail;
- order;
- language;
- form;
- channel.
If the grounds for an answer are insufficient, DyReP requests more information, makes the limitation visible, stops or refers the question to a person.
DyReP does not guarantee a correct AI answer. It organises the conditions under which professional work is carried out in a way that can be checked.
Worked example
Naming a source is not enough.
'Under normal use, we recommend inspection after twelve months. Under intensive use, the specialist assesses a shorter interval.'
'The source recommends inspection after twelve months under normal use. Under intensive use, a specialist has to assess whether a shorter interval is needed.'
'Inspection is needed every year.'
That last sentence loses the condition normal use and the role of the specialist.
A checkable answer therefore includes:
- the source and version used;
- relevant conditions;
- the distinction between what the source says and what has been inferred from it;
- the review status and who may decide.
Limit
This is an example written in advance. No AI model is running here and no real-world assessment has been recorded.