Switch AI tools without losing the story of your work.

A new tool is not the same conversation

You use one AI tool to organize your notes, then another to turn them into a checklist. The result looks polished. But can you still tell which details came from you, which were changed, and which version the second tool actually received?

Do not assume a second tool has the first conversation. Give it the context it needs deliberately. For a knowledge portfolio, the useful asset is not just the latest output. It is the work together with enough history to explain your choices. Moving between tools is a good moment to make that history visible.

Try a fictional workshop checklist

Imagine you are preparing a beginner's paper-craft workshop. Use made-up details for this exercise: a small group, a table of materials, and a sequence of simple activities. Write your own short notes first and save them separately. Do not use private participant information or someone else's unpublished materials.

Ask the first tool to organize those notes into a guide. Review its draft. Perhaps it adds a supply you did not choose or changes the order of activities. Correct the guide and save a reviewed version with a clear name, such as Workshop Guide v2. Keep the original notes as well.

Now ask a second tool to turn only that reviewed guide into a printable checklist. Explain that it should preserve the facts and flag any proposed additions. A formatting request should not quietly become permission to invent a new workshop.

Carry a small handoff note

Alongside the guide, include five plain-language details: the source you started from, the exact version being handed over, what you contributed, what the first tool helped do, and what the next tool is being asked to change. Add the date so the note has context.

For example: “Source: my fictional workshop notes. Input: Workshop Guide v2, reviewed today. My decisions: selected the materials and activity order. First tool: organized the wording. Next task: make a checklist without adding facts.” This is an original teaching template, not a technical standard.

After the second tool responds, compare the checklist with the reviewed guide. Did a required item disappear? Did an optional step become mandatory? Save your corrections and describe the checklist as derived from the guide. A new file name or a different tool does not erase the earlier work.

A record, not a guarantee

The W3C PROV Model Primer, published as a Working Group Note on April 30, 2013, explains provenance through entities, activities, and agents. It also describes derivation and revision relationships between entities. Those concepts offer a useful foundation for thinking about how one work product relates to another. The exercise here is a practical interpretation, not an implementation of PROV or a tested research result.

This connects AI governance with attribution: someone should be able to understand what was used, what changed, and who made the decisions. A handoff note can support that explanation, but it does not establish legal ownership, prevent copying, or guarantee that an AI tool follows instructions. Review the actual output.

Start with one harmless example today. Keep your original notes, your reviewed guide, the checklist, and the handoff note together. Your portfolio then shows more than the ability to generate a document. It shows how you carried your judgment across tools.