Knowledge / AEC-native data
Construction drawings
The 2D graphical record of design intent: plans, sections, elevations, details, schedules and title blocks, issued as sheets with revisions.
A drawing set is the most compact record of design intent a project has, and the hardest for software to read. Plans, sections, elevations, details and schedules carry meaning through symbols, references, revision clouds and title blocks, and the same element appears on several drawings at different scales. Teams navigate it by experience: knowing which drawing to open and which note matters. Structured knowledge turns the set into something a team can question directly, with every answer traceable to the drawing, the detail and the revision it came from.
What teams need to ask of it
Which drawings show this element, and do they agree with each other on size, level and location?
What is the current revision of each drawing, and what changed in the last issue?
Which details are referenced from this plan, and which references point to drawings that were never issued?
Where do the general notes or the legend override what the plan appears to show?
Which items on the schedule do not appear on any drawing, and which drawn items are missing from the schedule?
What does the drawing show that the specification does not cover?
Which drawings are affected by this RFI, this instruction or this variation?
Why generic AI gets it wrong
A drawing is not text with pictures. Most of what it says is carried by geometry, symbols and the relationships between drawings: a section marker on a plan, a detail callout pointing to another drawing, a revision cloud around a change, a key that assigns meaning to a hatch. A general-purpose model reads the text it can find on the drawing and produces a summary that sounds like the drawing without having understood it. It will name a door number it found in a tag and miss the note beside it that says omitted.
The other failure is scale. A set has hundreds of drawings and the answer to most questions lives on three of them, read together. A model that treats each drawing as a separate file cannot follow a reference from plan to detail to schedule, so it answers from whichever drawing it happened to read. The output is plausible, specific and unverifiable, which is worse than no answer at all.
One question, answered
QUESTION
ANSWER
SOURCES
We onboard a focused slice of your project data, configure the ingestion paths, extract a structured context graph, and build a workflow against the resulting structured knowledge layer.
Related to
Drawings answer to the specification, which states the requirements the drawings locate, and to the schedules, which list what the drawings show. The BIM model, where one exists, is the drawings' three-dimensional counterpart, and RFIs and variations are the record of what changed on them. Drawing coordination review and the specification-to-drawing consistency check both work across the set.
Does it read scanned or PDF drawings, or does it need CAD files?
What about revision clouds and revision notes?
Can it check the drawings against the model?
Related pages
Construction specifications
The written performance and prescriptive requirements for materials, workmanship and systems, organised by work section (NBS, CSI MasterFormat, Uniclass).
Drawing schedules (doors, windows, finishes, equipment)
Tabular schedules on drawings or in the model listing elements, their references, properties and locations, and the links between them and the plans.
IFC and BIM models
The federated 3D information model and its IFC exchange format: objects, properties, classifications, relationships and geometry.
Drawing coordination review
Checking architectural, structural and MEP sheets against each other for conflicts, missing references and inconsistent schedules before issue.
Specification to drawing consistency check
Reconciling what the specification requires with what the drawings and schedules show, item by item, with every mismatch traced to its clause and sheet.