Use the review queue

What the review queue is for

The review queue is not meant to be a busy dashboard. It is a focused approval workspace.

LeedBooks uses the queue when a transaction, rule, transfer, or cleanup item needs a human decision. The AI prepares the suggestion first. You approve it, adjust it, or reject it.

What you can review

The queue may show different kinds of work depending on your account activity and plan.

Common review items include:

• Uncategorized transactions.
• Suggested categories.
• Suggested transfer matches.
• Rule proposals.
• Vendor cleanup suggestions.
• Alias suggestions.
• Tax-readiness items.

Each item should answer one simple question: what decision does the business owner need to make?

Use AI when you are unsure

If a transaction does not make sense, ask the AI from the review flow.

Useful prompts include:

• "What is this vendor?"
• "Why did you suggest this category?"
• "Should this be a transfer?"
• "Have we seen this vendor before?"
• "Should I create a rule for this?"

The AI should help explain the decision, not force one. If the answer is still unclear, leave the item for later or ask your bookkeeper or tax professional.

Create rules carefully

Rules are powerful because they reduce repeated review.

Use a rule when:

• The vendor is recurring.
• The category is consistent.
• The description pattern is stable.
• The same decision would be correct next time.

Avoid rules when:

• The same vendor is used for different purposes.
• The amount or direction changes the meaning.
• The payment platform hides the real person being paid.
• You are only guessing.

Approve or reject

Approve items that are correct.

Reject items that are wrong or not useful. Rejection is part of the learning loop. It tells LeedBooks that the suggestion should not be repeated in that form.

If you edit a suggestion before approving it, the corrected approval is more valuable than a simple rejection because it teaches the system what the right answer should have been.

Keep it simple

The goal is not to make the user manage every accounting detail.

The goal is:

• The AI prepares the work.
• The user reviews only what matters.
• Correct approvals become future automation.
• The books stay current.

If the queue feels noisy, that is a signal to improve rules, aliases, transfer detection, or vendor cleanup.