A file organiser finds a document called Statement.pdf. The text mentions an account number, an address and a business name. Is it a household bill, a bank record, a customer document or something belonging to a project?
There may not be enough reliable evidence to decide.
That is where a good review process matters. “Review” should not mean the system has failed. It should mean the evidence is uncertain, conflicting or sensitive enough that a person should make the final choice.
This article explains a design principle for safe file organisation. It is not an announcement of a newly completed DataSortIQ capability.
Confident guessing is still guessing
Automatic sorting feels impressive when every file receives a destination. The problem is that a neat result can hide bad decisions.
A filename may be vague. A scanned page may be difficult to read. Metadata may be missing or misleading. One clue may point towards Finance while another points towards a customer project. If an organiser ignores the disagreement and chooses the strongest-looking label, the mistake can be difficult to notice later.
A safer system should be able to explain: “Here is what I found, here is what conflicts, and here is why I have not acted automatically.”
What belongs in Review?
Review is useful when:
- a file is unreadable, encrypted, damaged or in an unsupported format;
- two credible pieces of evidence point to different destinations;
- the proposed move could break a project or separate linked files;
- the file appears sensitive and the available context is incomplete;
- two similar files may be versions rather than exact duplicates; or
- a suggested name or destination would overwrite something already there.
These cases are not all equally urgent. A helpful queue should show the reason for review so people can handle high-impact decisions first instead of opening hundreds of unexplained files.
A review screen should help, not overwhelm
A useful review decision needs enough context to be understandable. That might include the current location, proposed category, relevant metadata, detected relationships and the evidence for and against the suggestion.
The person should be able to approve, change or leave the decision unresolved. “Leave it where it is” is a valid outcome. So is applying one carefully checked decision to a genuine group of related files.
What review should not become is a dumping ground for everything the organiser does not understand. Repeated uncertainty should inform future improvements: better format support, clearer rules or more useful evidence. Human decisions can guide design, but they should not silently become universal rules without testing.
A practical review habit today
You can apply the same idea without special software. When tidying files, create a temporary To Review area with a short reason in the name or a simple note: “possible duplicate”, “project link unknown” or “contains personal information”.
Review small groups at a time. Work from a backup, avoid deleting uncertain files, and record decisions that would be difficult to reconstruct. When you finish, check that important documents still open and that working projects can still find their supporting files.
How this relates to DataSortIQ
The current DataSortIQ Core version guide, checked on 2 October 2026, describes keeping uncertain or unsupported decisions under user control. It also states that collection, dependency and local semantic intelligence are being evaluated experimentally. Core remains in development.
The planned Image, Video and Audio / Music Studios remain Coming soon. Their specialist ideas require separate development, validation and release approval.
The aim of thoughtful automation is not to eliminate every question. It is to answer the safe ones clearly—and make the difficult ones easier for a person to resolve.