RIBIF - Intelligent Industrial Investment Platform

RIBIF is an end-to-end industrial project intelligence platform that turns a machine/material technical datasheet into a complete virtual factory package:

• structured technical extraction,
• 2D/3D plant layout,
• HD visualization video,
• global supplier sourcing,
• AI-assisted supplier negotiation (text + voice),
• TCO modeling,
• 5-year business plan + pitch deck,
• digital twin dashboard with operational KPIs,
• and a final technical catalog.

The system is implemented as one integrated web application with a Next.js frontend and Flask backend orchestrating the multi-module pipeline.

1) Global Objective

RIBIF is designed to support industrial companies from pre-engineering to investment decision by chaining 9 mandatory modules into one coherent workflow.

Input:

• Technical files (PDF, DOCX, and structured payloads)

Output:

• A virtualized factory solution with technical, procurement, financial, and operational decision support artifacts.

---

2) The 9 Modules (Functional + Implementation Mapping)

Module 1 - Tech Sheets & Needs Extraction

Business purpose:

• Ingest unstructured technical documentation.
• Extract key machine/material specs, constraints, capacities, utility requirements, safety factors.
• Build normalized stage-based JSON for downstream modules.

Backend implementation:

• Endpoint: POST /api/module1/process
• Status/result polling: GET /api/module1/status, GET /api/module1/result
• Route: backend/routes/module1.py
• Core logic: backend/utils/module_1/

Primary output:

• Structured extraction payload (stages 1..6) including stage_4_machines_equipements.

---

Module 2 - 2D/3D Factory Layout

Business purpose:

• Convert module 1 data into plant/workshop layout.
• Generate 2D and 3D artifacts representing zones, flows, and placement.

Backend implementation:

• Endpoint: POST /api/module2/layout
• Route: backend/routes/module2.py
• Pipeline engine: backend/utils/module_2/

Primary output:

• 2D layout artifact (DXF/PNG variants depending on flow)
• 3D model artifact (GLB/IFC-like workflow representation)

Notes:

• Layout route normalizes machine dimensions and can run with template fallback when skip flags are enabled.

---

Module 3 - HD Video Rendering

Business purpose:

• Transform 3D/layout visual context into investor-ready short video (site walkthrough style).

Backend implementation:

• Endpoint: POST /api/module3/video
• Route: backend/routes/module3.py
• Uses generated layout images or uploaded images + prompt-driven video generation.

Primary output:

• HD video file (.mp4) and downloadable URL.

---

Module 4 - Automated Supplier Sourcing

Business purpose:

• Identify supplier candidates globally.
• Aggregate price/lead-time/contact/evidence.
• Rank suppliers per machine family.

Backend implementation:

• Endpoint: POST /api/module4/enrich
• Route: backend/routes/module4.py
• Core agentic/bounded sourcing logic: backend/utils/module_4/
• Search providers and aggregation: backend/utils/web_search/

Primary output:

• machines_enriched[] list with supplier options and relevance scoring.

---

Module 5 - AI Negotiation (Text + Voice)

Business purpose:

• Support negotiation with suppliers through email-style text rounds and voice-interaction workflows.
• Produce negotiation history and best-offer summary.

Backend implementation:

• Core endpoints (subset):
  • POST /api/module5/text/negotiation
  • POST /api/module5/voice/negotiation
  • POST /api/module5/voice/summary
  • POST /api/module5/negotiations/start
  • POST /api/module5/v2/negotiations/create
• Route: backend/routes/module_5.py
• Negotiation logic: backend/utils/module_5/
• Email integration route: backend/routes/email_service.py

Frontend voice UI:

• frontend/src/app/voice-call/page.tsx
• Requires VAPI public key.

Primary output:

• Negotiation rounds, acceptance/rejection states, call transcript data, negotiated pricing traces.

---

Module 6 - TCO (Total Cost of Ownership)

Business purpose:

• Compute realistic industrial ownership cost over a horizon.
• Include purchase, installation, operations, maintenance, materials, and yearly projections.

Backend implementation:

• Endpoints:
  • POST /api/module6/tco-from-sourcing
  • POST /api/module6/tco
• Route: backend/routes/module6.py
• Core model: backend/utils/module_6/core.py

Primary output:

• Machine/product/raw-material TCO breakdowns
• Annual multi-year TCO projection.

---

Module 7 - Automatic Business Plan

Business purpose:

• Convert technical + TCO data into a 5-year financial model and investor narrative.
• Generate business-plan deliverables and pitch support.

Backend implementation:

• Endpoints:
  • POST /api/module7/business-plan
  • POST /api/module7/pitch-deck
• Route: backend/routes/module7.py
• Generators:
  • backend/utils/module_7/business_plan.py
  • backend/utils/module_7/pitch_deck.py

Primary output:

• Business plan file(s) and downloadable URLs
• Pitch deck output (including PDF export path when available).

---

Module 8 - Digital Twin & Digital Brain

Business purpose:

• Represent the factory as a live operational twin.
• Provide KPI dashboarding (OEE/TRS, throughput, availability, quality) with optimization-oriented visibility.

Current implementation in this repository:

• Primary UI route: frontend/src/app/twin/page.tsx
• Workflow twin view component: frontend/src/sections/DigitalTwinView.tsx
• 3D scene and KPI widgets are available from frontend modules.

---

Module 9 - Automatic Technical Catalog

Business purpose:

• Produce reusable technical documentation of machine families, product families, process chains, and logistics context.
• Deliver both document and web-consumable catalog outputs.

Backend implementation:

• Endpoint: POST /api/module9/catalog
• Route: backend/routes/module9.py
• Core generator: backend/utils/module_9/core.py

Primary output:

• Catalog payload + PDF artifact upload/URL.

---

3) Workflow Orchestration in Current App

The primary workflow engine is in frontend/src/app/workflow/[id]/page.tsx.

Actual execution order in the implemented pipeline:

1. Module 1 (spec extraction)
2. Module 4 (supplier sourcing)
3. Module 6 (TCO)
4. Module 7 (business plan + pitch deck)
5. Module 9 (technical catalog)
6. Module 2 (layout 2D/3D)
7. Module 3 (video)
8. Digital Twin view (Module 8) is available in UI, not as a blocking backend step.

Why this order:

• Financial and catalog decisions can proceed once sourcing and TCO are available.
• Layout/video are generated after core business/procurement intelligence has stabilized.

---

4) Repository Structure

ribif/
|- backend/
|  |- app.py                      # Flask entrypoint and blueprint registration
|  |- config.py                   # Paths, flags, env configuration
|  |- requirements.txt
|  |- routes/
|  |  |- module1.py
|  |  |- module2.py
|  |  |- module3.py
|  |  |- module4.py
|  |  |- module_5.py
|  |  |- module6.py
|  |  |- module7.py
|  |  |- module9.py
|  |  |- health.py
|  |  |- validation.py
|  |  |- email_service.py
|  |- utils/
|  |  |- module_1/
|  |  |- module_2/
|  |  |- module_4/
|  |  |- module_5/
|  |  |- module_6/
|  |  |- module_7/
|  |  |- module_9/
|  |  |- web_search/
|  |- template/                   # Template payloads for SKIP mode
|- frontend/
|  |- src/app/                    # Next.js App Router pages
|  |- src/sections/               # Workflow module views
|  |- src/lib/                    # API and Supabase helpers
|  |- supabase/                   # SQL schema/migrations

---

5) Technology Stack

Backend:

• Python + Flask
• LLM/tooling: OpenAI SDK, Google GenAI, LangChain/LangGraph
• Data processing: Pandas, NumPy, Pydantic
• Document extraction: PyMuPDF, python-docx
• CAD/render support: ezdxf, matplotlib, moviepy
• Persistence and file storage: Supabase

Frontend:

• Next.js (App Router) + React + TypeScript
• Tailwind CSS
• Three.js + React Three Fiber/Drei
• Recharts for KPI/dashboard charts
• Supabase JS client

---

6) Setup and Run

Prerequisites

• Python 3.10+
• Node.js 20+
• pnpm (recommended, lockfile included)
• Supabase project (recommended for full persistence)

Backend setup

cd backend
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
python app.py

Backend default URL:

• http://localhost:5000

Frontend setup

cd frontend
pnpm install
pnpm dev

Frontend default URL:

• http://localhost:3000

---

7) Environment Variables

Backend (backend/.env)

Core:

• GOOGLE_API_KEY (Gemini/Vertex-dependent steps)
• GEMINI_MODEL (optional override)
• IMAGE_MODEL (optional)
• VIDEO_MODEL (optional)

Supabase:

• SUPABASE_URL
• SUPABASE_SERVICE_ROLE_KEY
• SUPABASE_BUCKET_FILES (optional, default project-files)
• SUPABASE_BUCKET_PUBLIC (optional)

Search and sourcing:

• SERPER_API_KEY
• BING_API_KEY
• WEB_SEARCH_ENABLED
• PREFER_WEB_DATA

CORS and backend:

• BACKEND_BASE_URL
• CORS_ORIGINS

Optional module toggles:

• SKIP_MODULE_1 ... SKIP_MODULE_9
• CREATE_MACHINE_IMAGES
• CREATE_REALISTIC_RENDER
• CREATE_3D_MODEL

Module 5 negotiation/email:

• NEGOTIATION_SMTP_HOST, NEGOTIATION_SMTP_PORT, NEGOTIATION_SMTP_USERNAME, NEGOTIATION_SMTP_PASSWORD
• NEGOTIATION_IMAP_HOST, NEGOTIATION_IMAP_PORT
• TEST_RECIPIENT_EMAIL

Frontend (frontend/.env.local)

• NEXT_PUBLIC_BACKEND_URL
• NEXT_PUBLIC_SUPABASE_URL
• NEXT_PUBLIC_SUPABASE_ANON_KEY
• NEXT_PUBLIC_SUPABASE_SERVICE_ROLE_KEY (present in code; use with caution)
• NEXT_PUBLIC_VAPI_PUBLIC_KEY (voice call frontend)

---

8) API Surface (Core Module Endpoints)

Module endpoints:

• POST /api/module1/process
• GET /api/module1/status
• GET /api/module1/result
• POST /api/module2/layout
• POST /api/module3/video
• POST /api/module4/enrich
• POST /api/module5/text/negotiation
• POST /api/module5/voice/negotiation
• POST /api/module5/voice/summary
• POST /api/module6/tco-from-sourcing
• POST /api/module7/business-plan
• POST /api/module7/pitch-deck
• POST /api/module9/catalog

Supporting endpoints:

• GET /health
• GET /health/ready
• GET /health/live
• GET /health/metrics
• POST /api/validate-pdf
• POST /api/extract-text
• POST /api/email/send
• POST /api/email/poll

---

9) Data Model (Current Simplified Persistence)

The active schema migration (frontend/supabase/migrations/20260130_simplified.sql) uses a unified projects table as the central workflow state.

Key project columns:

• module1_data
• module4_data
• module5_data (added in 20260201_negotiations.sql)
• module6_data
• module7_business
• module7_pitch
• module9_data
• module2_data
• module3_data
• completed_modules (text array)
• settings, metadata

Negotiation tracking is stored in:

• negotiations table (frontend/supabase/migrations/20260201_negotiations.sql)

---

10) Artifacts and Deliverables

Typical generated outputs include:

• Module 1 extraction JSON
• Supplier shortlist/comparison JSON (Module 4)
• TCO report JSON (Module 6)
• Business plan + pitch assets (Module 7)
• Technical catalog PDF (Module 9)
• 2D layout and 3D model artifacts (Module 2)
• HD video .mp4 (Module 3)
• Digital twin dashboards/KPI views (Module 8 UI)

Sample deliverables are available under frontend/public/deliverables/.

---

11) Development and Verification

Frontend checks:

cd frontend
pnpm lint
pnpm test

Backend scripts/tests (examples in repo):

cd backend
python test_imports.py
python test_module4_budget_enforcement.py
python test_module1_refactor.py