Skip to content

Drawing sets

A hundred sheets, read as one set

A construction set is not a pile of floor plans. Sheets reference each other, schedules live apart from the views they describe, and the revision cloud on sheet A-203 changes what sheet A-101 meant.
  • Title block and sheet indexing
  • Plan views matched to schedules
  • Revision-to-revision comparison
Workflow diagram showing a construction drawing set being indexed, its callouts and notes extracted, and findings assembled for engineer review
Sheets indexed, then read according to what each one is.

Why it is hard

The information you want is split across sheets on purpose

Drawing sets are organised for the people who build from them, not for software. A door width lives on a schedule sheet; the door itself is a symbol three sheets earlier; whether it is fire rated may only appear in a note. Reading each sheet in isolation gets you fragments.
  • Sheet identity

    Title block read for sheet number, name, discipline, scale and revision, so each sheet is handled as what it is.
  • Schedule tables

    Grid structure recovered from ruled or unruled tables, then read row by row into records.
  • Cross-sheet links

    Tags on plan views matched to their rows on the schedules, surfacing what appears on one and not the other.
  • Revisions

    Two issues of a sheet aligned and compared, reported as elements added, removed and moved.

How it runs

Specialist models, chained

Each stage hands the next one something it can rely on. A language model only ever reasons over what has already been measured.
The full pipeline
  1. Stage 1

    Split and index the set

    Each sheet is separated, its title block read, and its type inferred — plan view, elevation, schedule, detail — so the right models run on it.
    Sheet index with numbers and types
  2. Stage 2

    Read every sheet in its own way

    Plan views go through detection and segmentation; schedule sheets go through table structure recovery and cell-level OCR.
    Per-sheet structured records
  3. Stage 3

    Reconcile across the set

    Tags detected on plans are matched to schedule rows by tag and by proximity, and mismatches are listed rather than resolved silently.
    Matched records and a discrepancy list
  4. Stage 4

    Compare against the previous issue

    Where an earlier revision exists, sheets are aligned and differences reported element by element instead of as a visual diff.
    Change list per sheet

What comes back

Data, with a trail back to the sheet

Delivered as JSON or CSV, with an annotated overlay so a person can check it. The exact fields are agreed with you before we build.
  • A sheet index with number, title, discipline and revision
  • Schedule rows as records, linked to the tags on the plan views
  • A discrepancy list where plan and schedule disagree
  • A per-sheet change list between two revisions
  • Overlays showing where every value was read

This is our base model — already trained on floor plans, not yet on yours. For a project it is fine-tuned on your drawings, against a target you set, on the fields you actually use. How fine-tuning works.

Start here

See it on a drawing of your own

Everything above is what our models do in general. What matters is what they do on your sheets, which takes one drawing and no meeting to find out.
  • One sheet is enough — PDF, DXF or a scan
  • Back in about 2 business days, with an annotated overlay
  • No call required, NDA on request

Send us one sheet

We run it through our floor plan models and send back the extracted data with an annotated overlay, so you can see what comes off your own drawings before anyone books a call.

  • Back in about 2 business days
  • PDF, DXF or a scan
  • NDA on request
This is our base model, before any tuning for you. On a paid project we train it further on your own drawings, which is where the rest of the accuracy comes from. So this shows you the starting point, not the best we can do.

By submitting, you agree to our Privacy Policy. We never share your details.

FAQ

Questions, answered

Questions this one comes with.
  • Sheets are processed independently and then consolidated, so set size is a question of runtime rather than feasibility. What actually limits things is consistency: a set drawn by one office over one period is far easier than a refurbishment package assembled from four sources across twenty years.

Get in touch

Tell us what is on your drawings

Describe the sheets you get and what you need off them. We will tell you what our models already handle, what needs fine-tuning on your drawings, and what is not worth automating at all.

Reply time
Within 1 business day

Would rather show than tell? Send one sheet for a free analysis

We reply within 1 business day.

By submitting, you agree to our Privacy Policy. We never share your details.