Tutorial ID | T-03 |
Section | Foundation |
Title | Choosing the Right Collection for Documents |
Subtitle | Place each file in the correct collection so the AI knows which records are real evidence. |
Before You Begin
You have a case with files in both the Documents collection and the AI Collections collection.
You know which files in your case came from outside the platform and which were generated on the platform, such as AI-generated reports or custom documents.
What You Will Accomplish
By the end of this tutorial, every file in your case will sit in the collection that matches where it came from. You will know how the Documents and AI Collections collections differ, and how to fix a file that ended up in the wrong one before it affects a workflow.
Why Does this Tutorial Matter
When a workflow runs, the AI draws only from the Documents collection — it does not look at anything stored in AI Collections. Keeping that separation correct means every workflow output reflects the real source records for the case, and nothing else.
Get it wrong, and the AI may treat its own past output as evidence — citing an AI-generated report as if it were an original record, and building new output on top of that circular reference. In a legal context, that risk can undermine the credibility of the entire case.
What to Do if Something Goes Wrong
The list below covers the most common problem with Documents and AI Collections. It follows the same pattern: what you'll notice, why it likely happened, a real example, and how to fix it.
Problem: An AI-generated report or custom document ends up in the Documents collection and gets treated as a source record.
Likely cause: An AI-generated report or custom document was moved from AI Collections into Documents — intentionally or by mistake — so the AI treats it as an external source in later workflow runs.
Example: A custom summary document is moved into the Documents collection. In the next workflow run, the AI cites that summary as if it were an original record, creating a circular reference where the AI's own prior output becomes evidence for new output.
Fix:
Open Files.
Open the Documents collection from the panel on the left and review its contents.
Identify any file in the Documents collection that was generated on the platform or created manually by a team member.
Open the editing mode for each file that needs to change and locate the collection type field.
Select AI Collections from the list to reassign the file to the correct collection.
Return to the Documents collection and confirm only external source records remain there.
Rerun any workflow that was connected to the incorrectly organized collection.
Tell the team that changing a document's collection type is a consequential action that must be communicated before it happens.
Tip Before running a workflow, do a quick pass through Documents and confirm nothing generated on the platform has landed there — it takes a minute and prevents a circular reference in the output. |
Summary
Every file in your case now sits in the collection that matches where it came from, so the AI draws only on genuine source records — and any workflow built on a misplaced file has been corrected.