How to track plan-review comments across resubmittals
A practical comment-log template and setup steps for tracking reviewer comments across rounds, with the parts AI can draft and the parts it can't.
Last checked 13 October 2026
A city review comes back. It might be a PDF letter, markups in a Bluebeam session, or the owner's own response form, like the MoDOT Review Comment Record Sheet. Before anyone writes an engineering answer, someone has to get every comment into a list. Then round two arrives with new numbering, and the list has to be reconciled.
A user on the Bluebeam community forum put it plainly: "We would like to keep track of these comments, along with any associated responses, throughout all submissions... Has anyone found the best way to keep track of these comments and were able to organize/tabulate these comments throughout these varying submission?" The replies offered workarounds, and none was marked as the answer. This post is the answer we would give.
What firms do today
- Slip-sheeting in Bluebeam carries markups onto the new sheet set. Forum replies point out that it breaks down when sheets are added.
- Exporting markups to Excel or XML and cleaning them up with Power Query. It works, but it needs someone who knows Power Query and a fresh cleanup each round.
- Retyping into a spreadsheet or the owner's form. Reliable, slow, and dependent on whoever did it last time.
- Keeping it in someone's head. Common, and the riskiest of the four.
A log that survives the third round
Whatever tool you use, the structure matters more than the software. Use one row per comment, with these columns:
| Column | Why it's there |
|---|---|
| Log ID (yours) | Stays the same across rounds; the reviewer's numbering will change |
| Round | 1, 2, 3 and so on |
| Reviewer or agency | Who sent it |
| Reviewer's comment number | As written in that round's letter |
| Discipline | Civil, mechanical, electrical, plumbing, landscape |
| Sheet(s) | As cited, plus the current sheet number if renumbered |
| Comment text | Word for word. Never paraphrase the reviewer |
| Response | Draft first, then final |
| Status | Open, drafted, engineer approved, sent, closed |
| Linked prior ID | If it repeats or continues an earlier comment |
| Responder and approver | Names and dates |
Rules that save pain later:
- Never renumber your own log IDs.
- Keep the reviewer's words exactly as written.
- Carry closed items forward with their status, so a repeated comment is easy to spot.
- Store the log in the project folder, next to each round's letter.
- Build the response letter from the log, not as a separate document.
Where AI helps
Most of this is text handling, which is where AI drafting fits well:
- Splitting a PDF letter or a markup export into one comment per row.
- Tagging discipline and sheet numbers from the comment text.
- Matching round-two comments to round-one rows, and flagging which are repeats with new numbers and which are new.
- Drafting a response for each comment in your firm's letter format or the owner's form, with anything that needs judgment flagged for an engineer instead of guessed.
Test each of these on letters where you already know the right answer before trusting them on a live project.
Where it doesn't
The engineering answer. Whether a pipe gets resized or a detail changes is a decision for the engineer in responsible charge. The boards that have written about AI say the same thing; see what the boards say about AI and the PE seal. Record who approved each response in the log itself.
What's coming from vendors
Bluebeam announced Bluebeam Max in October 2025. It adds natural-language prompts built on Anthropic's Claude that work over markup data, plus drawing-review features from Bluebeam's purchase of Firmus-AI. No comment-management feature was named, and pricing had not been published when we checked. If your reviewers send Bluebeam markups, watch it closely. Whatever you set up now should accept markup exports (CSV or XML files), not only PDF letters, so you're ready either way.
Other tools, such as CivilPlanAI, check plans before you submit them in order to reduce the number of comments. That is a different job from answering comments, and the two can work together. More in our tool comparison.
Setting it up, step by step
- Build the log template above in Excel or a SharePoint list.
- Pull your last 10 to 20 comment letters from the two or three jurisdictions you submit to most.
- Note each jurisdiction's habits: how it numbers comments, how it cites sheets, and whether it reuses numbers between rounds.
- Try AI splitting and tagging on one past letter where you already know the right result. Check every row against the original.
- Once splitting is reliable, add response drafting. Every response gets reviewed, and the approver goes in the log.
- Time the next live letter from receipt to sent, and compare it with how long letters took before. Our comment-letter cost worksheet shows how to turn those hours into dollars.
If step 6 shows no real saving, stop there. A clean log template on its own is still worth having.
A note on owner forms
Some owners require their own response format, as MoDOT does with its record sheet. Keep your internal log as the master and generate the owner's form from it. That way your history stays in one place even when every agency wants a different layout.
[Start a free conversation](/conversation?from=ai-for-engineering-firms and describe one recent comment letter. We'll walk through it with you and show what a log and a draft response would look like for your firm.