AI Native Design + ERP for Commercial Fit-outs

Project Brief: SpaceOS – AI-Native ERP for Commercial Office Fitouts 1. Executive Summary SpaceOS is a web-based software platform that combines three things into one seamless system: First, AI-powered design automation. When a floor plan changes, all related drawings update automatically. This includes AC ducts, fire alarms, thermostats, and wall elevations. Second, project management ERP. This covers budget tracking, procurement, scheduling, and subcontractor management. Third, client transparency portal. This provides real-time dashboards for clients to see progress. What already exists in the market: AI tools that automate MEP design exist. Examples include Endra and Augmenta. Fitout ERPs that manage budgets and schedules exist. Examples include Nexvia and FirstBit ERP. BIM clash detection tools exist. Navisworks is a good example. What does not exist is a single platform that does all three together with AI as the intelligence layer that connects them. SpaceOS will be the first to bring this to the market with an initial focus on the Middle East. This region has the highest global rates of project cost overruns and a strong appetite for AI adoption. 2. The Problem We Are Solving Problem 1: The Change Ripple Effect In commercial fitout projects, a small change like adding a new room to the floor plan creates a cascade of manual rework. When a room is added, the following must be updated manually: AC duct sizing and routing, new thermostat placement, additional fire sprinkler, new fire alarm device, wall elevations showing thermostat location, material quantities, budget, and schedule. Currently, engineers spend days or weeks manually updating each drawing across disciplines. This is slow, expensive, and error-prone. Problem 2: Fragmented Tools Project teams use separate tools for different tasks. Design happens in AutoCAD or Revit. Coordination happens in Navisworks. Budgeting happens in Excel. Communication happens through WhatsApp and email. This fragmentation creates data silos, delays, and miscommunication. Problem 3: No Single Source of Truth Clients, designers, project managers, and subcontractors all work from different versions of drawings and spreadsheets. This leads to costly rework and disputes. 3. The Solution: What We Are Building SpaceOS is a web-based platform with three integrated modules. Module 1: Design and Coordination Engine (Called Genesis) What it does: Accepts a floor plan upload in PDF or DWG format. Generates multiple optimized layout options using AI. When a user makes a change such as adding a room, the AI automatically performs several actions. It recalculates HVAC loads and re-routes AC ducts to the new room. It places a new thermostat on the nearest suitable wall. It adds a new fire sprinkler and fire alarm device in the new room. It updates all 2D drawings including floor plans, reflected ceiling plans, and wall elevations to show the new devices. It performs automated clash detection between the new MEP systems and existing structure. It allows users to review, modify, or approve changes before finalizing. Technical requirements for Module 1: The system must be built on a BIM engine like Autodesk Revit or Autodesk Platform Services as the single source of truth. The AI logic must include rule-based generative design for MEP systems including duct routing and device placement. It must also be customizable for regional building codes. Module 2: Project Management ERP (Called Orbit) What it does: AI-Powered Estimation: Automatically generates a Bill of Quantities and cost estimate from the approved design. Budget Tracking: Provides real-time budget versus actuals dashboard. Every purchase order and invoice updates the financials instantly. Smart Procurement: Auto-generates purchase orders and recommends vendors based on price, lead time, and quality history. Predictive Scheduling: Creates a project schedule that auto-adjusts when delays occur. Subcontractor Management: Tracks each trade's progress, quality, and payments. Change Order Intelligence: When a design change is made, the system instantly shows the cost and timeline impact. Technical requirements for Module 2: Must connect design data directly to budgets and procurement lists. Must support role-based access for project manager, designer, client, and subcontractor. Module 3: Intelligence Layer (Called Nexus) What it does: Anomaly Detection: Monitors costs, timelines, and progress to flag outliers before they become crises. For example, alerting when carpentry is 15 percent over budget. Predictive Risk Alerts: Provides warnings such as predicting a 72 percent chance of drywall delay and recommending reallocation of labor. Meeting-to-Action Engine: Records client meetings, generates minutes, extracts action items, and assigns them. Automatic Reporting: Generates weekly progress reports and client updates with one click. Post-Occupancy Analytics: Tracks how the space is used after handover to improve future designs. Technical requirements for Module 3: Must use machine learning models that improve with each project, creating a data network effect. Must integrate with the ERP and Design modules. 4. User Experience and Interfaces The platform will have three primary user interfaces based on user role. For the Designer and Engineer, they will see a Design Studio. Their key to know you fully ready this brief, start off your bid by telling me your favourite colour actions include uploading floor plans, generating layouts, modifying designs, reviewing AI-proposed MEP changes, and approving them. For the Project Manager, they will see a Project Dashboard. Their key actions include viewing real-time budget, schedule, and risks. They also manage procurement and subcontractors and generate reports. For the Client, they will see a Client Portal. Their key actions include viewing the layout and 3D walkthrough, seeing project progress, and approving change orders. Design Principles: Role-specific views must be enforced. Designers should never see procurement forms. Project managers should never see design tools unless they need to. Drag-and-drop simplicity for layout modifications. Mobile-responsive for site teams to upload progress photos and updates. 5. MVP Scope (Phase 1) For the initial Minimum Viable Product, we will build one complete change ripple workflow to demonstrate the platform's power. The MVP core workflow is Add a Room, then All Systems Update. Here is the step-by-step workflow: Step 1: User uploads a floor plan in PDF or DWG format. This requires a file parsing API. Step 2: AI generates 3 to 5 layout options. This requires an ML model for layout generation. Step 3: User selects a layout and adds a new room. This requires a web UI with drag-and-drop functionality. Step 4: AI recalculates HVAC load for the new room and proposes a new duct route. This requires a generative design algorithm. Step 5: AI places a thermostat on the nearest suitable wall. This requires a rule engine. Step 6: AI adds a fire sprinkler and fire alarm device. This requires a rule engine and BIM object library. Step 7: AI updates all affected 2D drawings including floor plan, reflected ceiling plan, and elevation. This requires BIM engine integration. Step 8: AI runs clash detection and flags any conflicts. This requires a clash detection engine. Step 9: Project manager reviews and approves the changes. This requires an approval workflow UI. Step 10: AI auto-generates updated Bill of Quantities, budget, and schedule. This requires ERP integration. MVP Exclusions for later phases: Full procurement automation will be added later. Post-occupancy analytics will be added later. Meeting-to-action engine will be added later. VR walkthroughs will be added later. 6. Technical Requirements and Stack Recommendations Required Architecture: For the frontend, use React and Three.js for web-based UI with 3D and 2D visualization. For the backend, use Node.js or Python with FastAPI for API endpoints and business logic. For the database, use PostgreSQL for relational data and a Vector DB for AI embeddings. For the BIM Engine, use Autodesk Revit API or Autodesk Platform Services as the single source of truth for all models. For AI and ML, use Python with PyTorch and LangChain for layout generation, anomaly detection, and generative design. For file storage, use AWS S3 or Azure Blob to store drawings, models, and documents. For real-time updates, use WebSockets for live collaboration and notifications. Integration Requirements: The BIM engine must be the central data repository. All changes must be made to the model first, and 2D drawings generated from it, not the other way around. The AI engine must be able to read the BIM model, propose changes, and write approved changes back to the BIM model. The ERP module must be able to read design data such as Bill of Quantities and material quantities and update budgets in real-time. 7. Development Team Requirements To build this platform, we need a team with specific expertise. We plan to outsource all development. Core Team Roles: BIM and Automation Developer: This person needs expertise in Revit API, C#, Python, Autodesk Platform Services, and Dynamo. Their key responsibilities are building the BIM integration, coordinating automation, and developing generative design algorithms. AI and ML Engineer: This person needs expertise in Python, PyTorch, Generative AI, LLMs, Computer Vision, and Rule Engines. Their key responsibilities are building layout generation, anomaly detection, risk prediction, and intelligent coordination. Full-Stack Developer: This person needs expertise in React, Node.js, PostgreSQL, WebSockets, and Three.js. Their key responsibilities are building the web interface, backend APIs, database, and real-time updates. UX and UI Designer: This person needs expertise in Figma, User Research, and Design Systems. Their key responsibilities are designing the role-specific interfaces and ensuring intuitive user flows. 8. Project Timeline Suggested Phase 0 is Discovery and will take 2 to 3 weeks. The deliverable is detailed technical specifications, architecture diagram, and tech stack finalization. Phase 1 is Core MVP and will take 3 to 4 months. The deliverable is a working MVP with the Add a Room to All Systems Update workflow. Phase 2 is Prototype Testing and will take 4 weeks. The deliverable is 5 design partners testing the MVP and providing feedback. Phase 3 is Public Beta and will take 2 months. The deliverable is the full ERP module with budget, procurement, and scheduling integrated. Phase 4 is Full Launch and is ongoing. The deliverable is post-launch features including VR, post-occupancy, and advanced AI. 9. Success Metrics At Month 6, we have the following targets. For time saved per design change, we aim to reduce manual rework from days to minutes. For user adoption, we aim to have 10 design partners actively using the platform. For client satisfaction, we aim for 90 percent reporting improved visibility and predictability. For accuracy, we aim for 95 percent of AI-generated MEP changes requiring no manual correction. 10. Why This Opportunity Exists Now The Middle East is the ideal launch market for several reasons. The region has the highest rates of project time and cost overruns globally. This creates urgent demand for better coordination tools. Regional firms like FirstBit ERP and Opteam have proven there is a market for fitout ERPs and AI project management. Global AI design companies like Endra from Sweden and Augmenta are expanding into the US and Europe but have no presence in the Middle East. This gives SpaceOS a valuable first-mover advantage. 11. Next Steps for Developers If you are interested in building SpaceOS, please respond with the following information. First, provide your team profile. Tell us who you are, your relevant experience especially in BIM, AEC, or AI software, and share case studies of similar projects. Second, provide your proposed tech stack. Explain how you would approach building the BIM-AI integration specifically. Third, provide a timeline and cost estimate based on the MVP scope above. Fourth, provide your questions. Tell us what information you need to refine your proposal. Appendix: Key Technologies and References Autodesk Platform Services (APS/Forge) for BIM data access and automation. Website: https://aps.autodesk.com Revit API for building custom Revit plugins. Website: https://revitapi.com VIKTOR for rapid prototyping of BIM web apps. Website: https://www.viktor.ai Endra as competitor reference for AI MEP design. Website: https://www.endra.ai Augmenta as competitor reference for AI MEP design. Website: https://www.augmenta.ai Three.js for 3D web visualization. Website: https://threejs.org React for frontend framework. Website: https://react.dev FastAPI for Python backend framework. Website: https://fastapi.tiangolo.com