Laydyne

What is Laydyne?

A browser workspace for planning rooms, workplaces, warehouses, clinics, and venues where an AI agent does the drafting. You connect ChatGPT, Claude, Codex, Claude Code, Cursor, or another MCP client to a single OAuth endpoint, keep the Studio tab open, and ask the agent to build a measured plan from a drawing, a photo with one known dimension, or a written brief. The model shows as a 2D plan, a 2.5D view with heights, and a 3D perspective, and free route checks flag overlaps, narrow passages, and unreachable targets. Paid plans add discrete-event simulations of people, robots, and goods so you can compare layout and operating options against stated objectives.

Record written 2026-10-07 from the vendor's submission and site. The site is offered in English, Japanese, Chinese, Spanish, German, French, Portuguese, and Korean, and the copyright line reads 2026 Laydyne with no company details.

Key strengths

• One MCP endpoint (laydyne.com/api/mcp/oauth) with OAuth and automatic client registration connects ChatGPT, Claude, Gemini, GitHub Copilot, Codex, Claude Code, Cursor, and Mistral Le Chat, and WebMCP lets a browser-based agent call it directly
• imports PNG, JPEG, WebP, and PDF files up to 5 MB for AI-assisted drafting, including photo reconstruction from one measured dimension
• exports a portable JSON project file, scaled SVG or PNG floor drawings with dimensions and zone data, and PNG 3D images
• route checks test overlaps, narrow passages, and reachability for free, and clearance checks compare against declared minimums
• simulation tools cover single-floor cleaning or two-station transport up to 60 by 60 m, cleaning robot fleets of 1 to 8 with cost inputs, and multi-floor arrival and service scenarios averaged over 10 runs with recorded playback
• a public library of 32 example spaces, a Claude Code skill file, and a dependency-free JavaScript kit are published

Common use cases

Limitations to know

The vendor states that the simulation is a simplified deterministic model in which robots are circles moving at constant speed, with no collision avoidance, crowd physics, or optimised lift control, and calls the results a way to explore options rather than a calibrated prediction. It also says it cannot verify that your drawing, object sizes, or operating assumptions match the real site. There is no DXF, DWG, or IFC import or export, so drawings leave as SVG, PNG, or JSON. Projects save to browser storage by default, cloud saving is limited by plan, and the Studio tab must stay open for the MCP connection to work. No company location or registration is published. AEC Hub has not tested the product.