Vela CAD · In development
Vela CAD
The CAD copilot
you talk to.
It runs on top of the CAD you already use — reading your drawing and drawing alongside you. Tell it what you want, and the drawing on screen actually gets tidied. It reads coordinates, not images, so it never guesses.
Still early. This isn't a finished product yet — it's the direction we're heading.
The problem
Half your time in CAD isn't design
Tidying layers, aligning dimensions, normalizing annotations, reworking someone else's drawing to your office standard. The part that's really “design” is only a slice. The rest is manual repetition, every day. Vela aims to take that repetition off your hands.
Today
- You tidy layers and annotations by hand, every time you open a drawing.
- You re-organize received drawings to your office standard from scratch.
- Even a simple edit means running several commands in sequence.
With Vela
- ✓ “Clean up this drawing” — one sentence, and it’s tidied.
- ✓ Say what you intend in plain words; the AI does it on screen.
- ✓ Vela handles the repetition. You keep the judgment.
Why it's different
It reads coordinates, not pictures
Most AI looks at a drawing like a photograph. It sees the pixels on screen and guesses — “that line is probably a wall.” Guesses can be wrong. A line off by 0.1 mm, two lines slightly overlapping, a dimension a hair out of place — a photo blurs those differences.
Vela reads the real data the drawing holds: the exact start and end coordinate of every line, every layer, every block. Not the picture, but the design data that made the picture. So it doesn't guess — it knows what's there, and exactly where.
Looks at the on-screen pixels and guesses.
Reads the drawing’s coordinate and geometry data directly.
Smears over overlapping or slightly-off lines.
Catches differences down to 0.1 mm, as written in the data.
Redraws based on what it saw.
Edits the original design data directly.
The engine
Vector Insights
The heart of Vela CAD. A recognition engine that treats a drawing's coordinates, layers, and block structure as primary data. “Insight drawn from vectors” — exactly as the name says, it reads the design data instead of interpreting a picture.
Coordinate-based recognition
It grasps lines, arcs, and blocks as geometry, not pixels — reading the numbers behind the shape, not the shape itself.
Structural understanding
It reads how layers and blocks relate, and how the drawing is put together — understanding what each line means in context.
Real-time tracking
As the screen changes, recognition moves with it. Not one frozen frame, but the drawing you’re working on right now.
Engine generation
Current generation — Carina 3.x
We name engine generations after constellations. Vela (the Sails) and Carina (the Keel) were both part of the old constellation Argo Navis — the ship that sailed. A fitting name, we think, for a copilot that helps you navigate a drawing.
In use
Soon — the real screen
Demo · coming soon
We'll show the moment a single sentence tidies a drawing — on video, soon.
Early access
See it first
Vela CAD is still being built. Join the list and we'll tell you the moment an early build is ready. If you share which CAD you use and what's most tedious about your day, it genuinely helps us aim.
What you send is used only to shape the product. No marketing email.
Join early access
One email is enough. Add a line about your CAD and your most tedious task, and we'll keep it in mind when the early build is ready.
Email [email protected] →You can also use the contact page. Vela CAD is Sector Zero Studio's second product — feel free to look at AutoPlotter first.