Design quality review

A design quality review looks for conflicts within the plans, conflicts within the specs, and clashes between the two.

1. Type / in any chat and choose Design Review. The agent first proposes an outline: a high-level review strategy; architectural, structural, civil, and mechanical review; cross-discipline coordination; and constructability and bidability.
2. Scope or change the outline — focus on a particular area (say, concrete only), drop disciplines you don't need, use the @ symbol to point the agent at specific sections, or attach extra files. You can also tell it to check against publicly available standards like IBC, or to flag zoning and code conflicts. Then hit Approve.
3. The agent returns an overall conclusion plus issues. Each issue is independently verified against the documents before it lands in the project's Issues tab, so the list you triage is already filtered for substance.
4. Keep engaging with the agent: ask it to write up an RFI for the issues it found, ask about HVAC means, rebar sizes, quantity counts, or measurements — and put results into tables.

A review runs on its own — a regular review takes roughly 20–30 minutes, a deep review up to a couple of hours. You can close the window while it works, and run multiple projects in parallel instead of waiting for one to finish.

• Verified before you see it — A second, independent AI checks every finding by trying to disprove it against the plans and specs. Confirmed issues arrive with the evidence attached; findings that do not hold up are archived automatically with the reason, and minor self-resolving items go to a Cleanup list instead of your queue.
• Custom workflows — Turn your QA/QC checklists and design standards into standardized, one-click review agents — for example, separate commercial and residential reviews.
• Rework cost exposure — Every issue carries a cost estimate: a Rework figure when the documents guarantee field rework, a Risk figure for latent failure modes, or a Resolution figure — what it costs to resolve the issue in the documents before construction. Issues roll up to a dashboard of rework cost exposure, with Risk and Resolution tracked alongside the headline. Lists can be sorted, filtered, and assigned out by discipline or assignee.
• Side-by-side review — Issue description and a suggested resolution on the right; the AI auto-flags the markup on the sheet at left, with multiple reference sheets. Confirm each finding, or mark it invalid, trivial, or incomplete — the agent learns what matters to your organization over time.
• Built for multiplayer — Run a review and every member on the job can engage and collaborate on the issues together, all at once.
• Deep design quality review — A regular review works from the agent's own judgment; a deep review forces it through every spec section against the plans — plus each document's internal consistency and the clash between them — surfacing far more issues. Complex projects run in the background for a couple of hours.