ENR reports on Structured AI's $4.2m fundraise and partnership with Syska Hennessy

Revit

Reviews that never leave Revit.

No export. No re-entry. Structured AI checks the real elements in your Revit model and surfaces every finding where it was made, leaving the judgment to your reviewers.

The Revit
Workflow

QC where your drawings are born

Currently for many firms, design happens in Revit, but reviews happen on exported PDFs: issues get clouded in Bluebeam, emailed around, and re-entered into the model. Every revision repeats the cycle of export, markup, and manual fix. Structured AI runs its checks on the model’s actual elements inside Revit and surfaces every finding where it was made, so the errors reach your reviewers instead of the field. Nothing changes in the model until someone applies it.

Today, every engineer and architect without Structured AI works like this

Structured AI changes this

Structured AI finds the errors right inside Revit and puts each one in front of you to approve, before your drawings ever go out the door.

50%

of billable hours at a typical engineering firm go to checking drawings

3 in 10

errors are caught by a manual review pass. The rest surface in the field

500+

sheets in a modern set, re-exported and re-checked at every revision

Other tools scan pixels on a PDF.
Structured AI reads the live data in your Revit model.

Tools that work on exported PDFs are interpreting pixels: lines, symbols, OCR. They can flag that something might be wrong. Structured AI’s agent runs inside Revit and writes native code against the Revit API, so it works with the model’s actual objects: every parameter, tag, and schedule, across your linked architectural, structural, MEP, and civil models.

The difference shows up in the results. A PDF check on this plan reports a possible gap to verify by hand. The model check traverses the duct network itself and returns a definitive answer, tied to the exact element.

Watch it work the model

The add-in puts the full platform inside Revit: the same check library, code compliance, and custom checks, all run against live geometry. Every result is deterministic (pass, fail, or verify) and links to its element, so a click takes you there.

When a fix is approved, it lands in the model: spelling, tags, schedule values, revision clouds. Then the clean set exports itself. In one client run, the agent traversed 5,000 HVAC connectors in 30 seconds and found multiple disconnected ducts.

Ran 6 checks 12
L2 Mechanical linked: Arch, Elec
  • A-114 Egress widths: 214 doors verified
  • A-091 Fire-rated walls: tags match schedule
  • E-027 Panel L2-B: plan tag differs from single-line diagram 1
  • M-042 Duct 4-117: undersized for 850 CFM 1
  • A-133 ADA clearances: 96 rooms pass
  • M-008 Disconnected ducts: 10 across 5,000 connectors 10

6/6 done · 0 running · 0 pending · elapsed 0m 30s

Structured AI is built for the whole team

Today

Finds out three weeks after export, via a senior’s markup, that two ducts never got connected.

With Structured AI

Checks run in the model as they work. Each finding navigates straight to the element, so the exported set is already clean.