I've been poking at AI tools that produce useful output (not just pretty pictures), and this week's find genuinely fit that bar: Floor Planner, an AI floor planner that turns text descriptions into dimensioned floor plans — and a companion image to DWG converter that turns scanned architectural drawings into editable CAD files.
I ran both through a real task: re-planning our apartment layout and digitizing an old paper plan. Here's the workflow.
The generator: describe it, don't draw it
The floor planner has three input modes:
Mode
What it does
Room
Single-space design (one bedroom, one kitchen)
House
Full layout — set room counts, kitchen type, floors, total area
Custom
Free-text prompt, e.g. "compact 2-bedroom apartment, 58 m², open kitchen, laundry closet"
You set dimensions in m² or ft², choose one of 10 rendering styles (Technical, Blueprint, 3D Isometric, Realistic, Hand-Drawn, Watercolor, Marker Pen, Colored 2D, ...), and hit Generate. A high-res plan comes back in seconds — as 2D, 2.5D or 3D. Results auto-save to a History tab, and you can change dimensions or switch styles without starting over.
Two things stood out from a workflow perspective:

  1. Style is decoupled from layout. The same plan renders in 10 styles, so one floor planner output becomes ten artifacts — Realistic for stakeholders, Technical for contractors, Hand-Drawn for concept decks.
  2. Image + prompt input. You can upload a photo of the existing space and prompt for changes. The floor planner reinterprets the image into a new plan, which is exactly the ergonomics you want for renovation work where "current state" is the starting point. The converter: image to DWG The second page handles the opposite direction — legacy assets. If you have a scanned or photographed architectural drawing (JPG/PNG), the image to DWG tool converts it into DWG, the format AutoCAD and most CAD pipelines use. I fed it a 20-year-old paper plan I'd photographed with a phone. Conversion produced a DWG that opened in a CAD viewer with walls and annotations as vector objects — selectable and editable, not a flat image. Why this matters in practice: scanned plans are usually dead ends in a digital pipeline. You can view them, but not modify them. An image to DWG step converts an archive into an asset. And it composes nicely with the generator — old plan goes in one end as an image, comes out the other end as DWG, then the AI floor planner can iterate new versions of it from a prompt. The combined pipeline paper plan → photo (JPG/PNG) → image to DWG (editable CAD file) → iterate with AI floor planner (prompt → 2D/2.5D/3D) For anyone building products around renovation, real estate, or interior design, this is a cheap way to prototype the "before → after" experience. No sign-up to try the floor planner, no CAD knowledge required, and the output is presentable enough for early-stage client conversations. Verdict Not a CAD replacement — it's a front-end for the ideation phase that used to cost hours. The floor planner handles generation; the image to DWG converter handles digitization. Together they cover both halves of a renovation workflow. Links:
  3. AI Floor Planner: https://floor-planner.ai/
  4. Image to DWG converter: https://floor-planner.ai/image-to-dwg