Skip to main content
The QC Analysis workflow lets you upload construction drawing sets as PDFs and run a two-phase review: first, a set of deterministic checks that cross-reference callouts and schedule tags across sheets; second, an AI agent that reviews each sheet individually and flags coordination issues, missing information, and potential errors. Findings appear as they are generated so you can start reviewing before the full analysis completes.

​
Starting a new analysis

1

Select a project

Choose the project from the sidebar. QC runs are scoped to a project, and the upload form will prompt you to select one if none is active.
2

Upload drawing PDFs

Drag your PDF drawing set onto the upload area or click to browse. You can upload multiple PDFs in a single run. Each file is processed as a separate document.
3

Choose checks to run

Select which automated checks to include. At minimum you can enable:
  • Callout cross-reference — verifies that every detail callout on a sheet points to a sheet and detail that actually exist in the set.
  • Schedule tag reconciliation — checks that every tag on a floor plan or elevation has a corresponding entry in the relevant schedule.
You can also enable the AI agent sheet review for a deeper per-sheet analysis.
4

Start the analysis

Click Start Analysis. The analysis timeline appears immediately and tracks progress through four stages.

​
Analysis stages

The progress timeline shows four sequential stages. The active stage displays a live substatus beneath its label.
The AI sheet review stage starts after the deterministic checks finish, but the two phases are independent. You can begin reviewing tag and callout findings while the agent is still running.

​
Viewing findings

Once results are available, findings are displayed in a list that you can filter and sort. Each finding includes:
  • Sheet number and page reference
  • Title — a short description of the issue
  • Details — the full explanation and any source references
  • Severity — one of three levels:
    • informational — worth noting, no immediate action required
    • warning — a potential problem that should be reviewed
    • error — a definite issue that must be resolved
  • Source — either automated-check (deterministic rule) or agent (AI review)

​
Callout cross-reference findings

These findings flag callouts where the referenced detail number and target sheet combination cannot be found elsewhere in the drawing set. Each finding shows the callout text and the page it was found on.

​
Schedule tag reconciliation findings

These findings flag tags that appear on sheets but have no corresponding row in the expected schedule. The finding title shows the tag identifier and, where available, the schedule it was checked against.

​
AI agent sheet review findings

Agent findings are generated by the senior-reviewer AI after it reads each sheet in context. These findings cover a wider range of issues than the deterministic checks — coordination gaps, unclear dimensions, missing references, and specification conflicts.

​
Managing findings

Use the filter controls above the findings list to narrow results by:
  • Category — callout-cross-ref, schedule-tag-reconciliation, or agent findings
  • Severity — informational, warning, or error
  • Sheet — limit to a specific sheet number
Filter to error severity first and work through those before reviewing warnings. This keeps your attention on the findings most likely to cause RFIs or rework.

​
Run history

The run history sidebar on the right lists every previous analysis for the current project. Each entry shows the run date and a summary of finding counts. Click any past run to load its findings. The main panel switches to that run’s results without affecting your current analysis. This lets you compare what changed between drawing revisions.
Switching to a past run does not delete the current run. You can return to it at any time by selecting it from the history sidebar.

​
Re-running analysis

To analyze a new revision of the drawing set, click New Analysis and upload the updated PDFs. Each submission creates a separate run entry in the history sidebar, so you retain a full audit trail of findings across all drawing revisions.