Skip to content
← Proof
Proof · Case

Every review comment gets its governing drawing attached

Daewoo E&C · DW-SIGHT

Building committee review comments and action plans are parsed automatically and matched one-to-one against the drawing registry; every match carries a confidence level and its evidence, and a person makes the final call. On a live project, all 13 of 13 review comments were confirmed as reflected — in use on real sites including Cheongna and Gyeongbuk.

500 sheets1:1 matched

basis: 13 of 13 items on a live Daewoo E&C DW-SIGHT project

Context — background

Passing a building committee review means confirming that every single review comment has been reflected in the drawings. The comments are prose and the drawing set runs to hundreds of sheets — the reconciliation depended entirely on one person's memory and eyes.

The existing workflow and its bottleneck

Manually linking each review comment and action plan one-to-one to the drawing registry — whether an omission was caught depended on the diligence of the individual reviewer.

Data supplied

Review comment sheets and action plans, plus the project drawing set (several hundred sheets).

doAZ capabilities used

Each review comment is matched against the drawing registry and judged as reflected, partially reflected or not reflected, with the governing drawing and its location attached to the verdict. Consistency checks — such as a full reconciliation of the window and door schedule against the floor plans — run alongside.

What the AI does

Interpreting the meaning of each comment and proposing candidate drawing matches.

What rules, calculation and verification do

Quantitative determinations such as schedule reconciliation and instance counting are performed by deterministic code — when match confidence is low nothing is confirmed, and the item is handed to a reviewer instead.

Where a person reviews

The reviewer confirms every match and verdict on the drawing itself before it is finalized.

Measured outcome and its conditions

On a live project of roughly 500 sheets, all 13 of 13 review comments were traced end to end for reflection in the drawings

Basis: 13 of 13 items on a live Daewoo E&C project.

Known limitations

On sheet sets where drawing titles and annotations are non-standard, automatic match confidence drops and a larger share of the work goes to human review — in those cases the system does not fill in a plausible answer; it surfaces the item as pending review.

Customer approval scope

Customer name and item counts approved for publication.

We run the same verification on your own material

We agree on the accuracy definition against a frozen gold set and then measure it — a structure in which cherry-picking is impossible.

Request a technical verification demo