AI Agent That Designs Building Systems Instantly
Python Scripts
Mepit AI is an intelligent automation pipeline that instantly transforms raw 2D architectural floor plans into fully compliant fire protection system layouts, eliminating the need for manual CAD drafting. By combining Computer Vision, Large Language Models, and geometric graph theory, it reads structural boundaries and semantic labels directly from DXF/DWG files to automatically place sprinklers, Manual Call Points, and hooters with precision-engineered logic rather than generic proximity guesses. The platform calculates fire travel distances against 30m/45m code limits, models acoustic drop-off to verify 65dB audibility targets, and routes sprinkler piping with collision-free, orthogonal alignment — all while enforcing strict clearance rules like a 200mm minimum wall distance.
AI Agent That Designs Building Systems Instantly
Mepit AI is a proprietary software pipeline that fully automates the placement and routing of fire protection systems on architectural floor plans. Built for MEP engineering firms, architects, and fire safety contractors, it replaces manual CAD drafting by ingesting raw 2D floor plans and automatically generating compliant hardware layouts.
By combining Large Language Models with Computer Vision and geometric graph theory, Mepit AI reads structural boundaries and semantic labels directly from DXF/DWG files to accurately plot sprinklers, Manual Call Points (MCPs), and hooters. It functions as an intelligent automation layer that instantly maps safety requirements — including 30m/45m fire travel distance limits and 65dB acoustic targets — cutting hundreds of hours of manual drafting per project while maintaining strict adherence to international fire safety codes.
Built on Python and the ODA CAD framework, Mepit AI gives MEP engineering firms, architects, and fire safety contractors an AI-powered layer that turns hours of manual layout work into an automated, code-compliant process.
Core Capabilities
- Intelligent Floor Plan Parsing
Combines CV-based semantic masking (walls, doors, and structural elements identified by color/class) with LLM API integration to classify DXF/DWG text labels extracted through the ODA (Open Design Alliance) framework — turning raw architectural drawings into structured, machine-readable floor plan data. - Precision Hardware Logic
Goes beyond generic proximity-based search radii. Door-adjacent hardware placement uses precise label-based logic tied directly to door boxes, and NLP tagging is applied to granular geometric sub-boxes rather than whole room polygons — producing highly accurate, code-defensible hardware placement instead of approximate "nearest wall" guesses. - Advanced Geometric Routing & Pathfinding
Uses convex polygon decomposition to spatially map each floor plan, then mathematically calculates fire travel distances and models acoustic drop-off through obstacles (e.g., a 20dB reduction through closed doors) to validate hooter/alarm audibility across the building. - Collision-Free Pipe Alignment
Automated pipe routing enforces strict global alignment rules, including a fixed 200mm minimum wall clearance for sprinklers. Topological routing logic guarantees clean, orthogonal layouts and eliminates common CAD-automation glitches like inclined pipe segments at tee-junctions.
Who It's For
- MEP Engineering Firms — automate first-pass fire protection layouts across large project volumes
- Architects — validate fire safety compliance early in the design process without waiting on MEP consultants
- Fire Safety Contractors — generate accurate, code-aligned hardware placement and routing for bids and installations
- CAD/BIM Service Providers — offer automated fire protection drafting as a value-added service
Tech Stack
| Layer | Technology |
|---|---|
| Core Logic & Engine | Python |
| CAD Interoperability | ODA (Open Design Alliance) platform, open-source CAD parsing libraries |
| Computer Vision | Semantic segmentation for wall/door/structural classification |
| AI/LLM | LLM API integration for text and label classification |
| Geometry & Routing | Custom spatial algorithms, convex polygon decomposition, graph theory pathfinding |
| File Formats | DXF, DWG |
FAQ
1. Does Mepit AI replace a licensed fire protection engineer?
No. Mepit AI automates layout generation and code-distance calculations to dramatically speed up drafting, but final designs should still be reviewed and stamped by a licensed engineer where required.
2. What file formats does it accept?
Raw 2D floor plans in DXF or DWG format, processed through the ODA platform for precise drawing data extraction.
3. How does it handle acoustic compliance for hooters/alarms?
It models acoustic drop-off mathematically — including a 20dB reduction through closed doors — to verify that 65dB audibility targets are met throughout the floor plan, not just at the source device.
4. Why not use simple proximity radii for hardware placement?
Generic proximity-based logic produces false positives near doors, corridors, and irregular room shapes. Mepit AI uses label-based logic tied to actual door boxes and granular sub-box NLP tagging for placement that holds up under code review.
5. Is the source code unencrypted?
Yes — full unencrypted Python source code, one-time license, no domain restrictions, white-label ready.
Don’t Keep This to Yourself — Share It!
License
A regular license permits the use of an item in a single project, whether for personal or commercial purposes, by you or on behalf of a client. Resale of the item, either individually or as part of a project, is prohibited, and the distribution of source files is not allowed.
Server Requirements
- Core Logic & Engine: Python
- CAD Interoperability: ODA (Open Design Alliance) platform, open-source CAD parsing libraries
- Computer Vision: Semantic segmentation for wall/door/structural classification
- AI/LLM: LLM API integration for text and label classification
- Geometry & Routing: Custom spatial algorithms, convex polygon decomposition, graph theory pathfinding
- File Formats: DXF, DWG
Payment Type
One time Payment: Upon receiving your full payment, we will provide you with the complete source code. After that, no additional payments will be required.
Delivery Time
We will send the Download Link to you Within 2-4 hrs. On Working Days(Monday- Friday) And the source code will deliver within 4-8 hours On Weekends (Saturday -Sunday).
Technical Support
Unlimited Email Support for One Year.
Script Update
Free Updates for One Year.
Customization Scope
Includes Complete Source Code.
Encryption
No Encryption and No Domain Limitations.
Installation
No Free Installation Services. Our charges range from $5 to $10, depending on the software's complexity and criticality.
No comments yet. Be the first to comment!