What do [VERIFY] and [REVIEWER] mean in the corrections letter?
Quick answer: [VERIFY] items need a reviewer's visual confirmation on the sheet — the AI thinks something may be missing but isn't certain. [REVIEWER] items need a reviewer's professional judgment — the AI can't decide without human expertise. Both show up inline in the corrections letter as a tag at the start of the item.
The three tag levels in practice
The generation footer at the bottom of the letter summarizes the breakdown — e.g., "23 correction items (13 CONFIRM, 5 VERIFY, 5 advisory)":
| Tag | What it means | What you do |
|---|---|---|
| (no tag / CONFIRM) | AI is confident — clear miss or violation. | Accept or Modify on the Findings tab; it goes into the letter when accepted or modified. |
| [VERIFY] | AI suspects a miss but can't be sure without closer inspection. | Open the sheet, look at the location the AI cited, then Accept, Modify, or Reject it. |
| [REVIEWER] | Needs a licensed reviewer's judgment (e.g. fire rating at an ambiguous setback). | Apply your professional judgment. Accept if you agree, Modify if you want to reword, or Reject if it should not be a correction. |
Example (from the Buena Park / Placentia sample)
- "7. [VERIFY] Add Placentia Municipal Code references to governing codes list. The cover sheet governing codes list should reference applicable PMC chapters… Reviewer should verify whether PMC references are already present on the cover sheet."
- "11. [VERIFY] Identify fire-rated exterior wall assemblies. At the proposed 4-foot setback, exterior walls of the ADU are within 5 feet of the property line, triggering a 1-hour fire-resistance rating requirement per CBC Table 602…"
Both of those are items where the AI wants you to eyeball the sheet before it becomes a formal correction.
Why the tags exist
The letter is explicit that "All items must be verified by a licensed plan checker before issuance." The tags make it obvious which items need extra human attention. They are a safety mechanism, not a limitation — without them, the AI would either over-call (noisy letter) or under-call (miss real issues).