wecan-innosuisse-ai-draft

Draft documentation for wecan-innosuisse-ai (Innosuisse 133.672 IP-SBM)

Information

Property	Value
Language	HTML
Stars	0
Forks	0
Watchers	0
Open Issues	0
License	No License
Created	2026-03-11
Last Updated	2026-03-11
Last Push	2026-03-11
Contributors	1
Default Branch	main
Visibility	private

Datasets

This repository includes 30 dataset(s):

Dataset	Format	Size

| data | | 0.0 KB |

| extracted_content.json | .json | 558.35 KB |

| progress_snapshot.json | .json | 69.26 KB |

| project_plan_data.json | .json | 46.45 KB |

| review_findings.json | .json | 1.95 KB |

| M01_January2026.json | .json | 16.55 KB |

| M02_February2026.json | .json | 13.33 KB |

| M03_March2026.json | .json | 11.6 KB |

| M04_April2026.json | .json | 15.38 KB |

| M05_May2026.json | .json | 12.27 KB |

| M06_June2026.json | .json | 18.35 KB |

| M07_July2026.json | .json | 9.53 KB |

| M08_August2026.json | .json | 9.54 KB |

| M09_September2026.json | .json | 14.06 KB |

| M10_October2026.json | .json | 13.42 KB |

| M11_November2026.json | .json | 14.75 KB |

| M12_December2026.json | .json | 24.13 KB |

| M13_January2027.json | .json | 11.37 KB |

| M14_February2027.json | .json | 14.0 KB |

| M15_March2027.json | .json | 17.54 KB |

| M16_April2027.json | .json | 14.66 KB |

| M17_May2027.json | .json | 10.04 KB |

| M18_June2027.json | .json | 10.58 KB |

| M19_July2027.json | .json | 12.69 KB |

| M20_August2027.json | .json | 16.36 KB |

| M21_September2027.json | .json | 14.18 KB |

| M22_October2027.json | .json | 8.65 KB |

| M23_November2027.json | .json | 8.16 KB |

| M24_December2027.json | .json | 17.17 KB |

| schema.json | .json | 8.56 KB |

Reproducibility

No specific reproducibility files found.

Status

Issues: Enabled
Wiki: Enabled
Pages: Enabled

README

AI Orchestrator

A Specialized and Secure AI Orchestrator for Swiss Financial Compliance

Project Overview

This Innosuisse Innovation Project develops an on-premise AI system that automates document processing for Swiss financial institutions while ensuring data sovereignty. The solution combines OCR with large language models to transform manual document handling into automated workflows.

Application: 133.672 IP-SBM Duration: 24 months (January 2026 - December 2027) Total Funding: CHF 591,240 (Innosuisse: CHF 327,336)

Key Capabilities

Capability	Description	Target
Document Digitization	OCR + LLM extraction from complex scanned documents	50-100 page documents
Dynamic Field Mapping	AI-driven mapping to target systems without manual configuration	50+ enterprise schemas
Document Pre-filling	Automated population of PDFs/Excel with variable structures	90% accuracy

Objectives

OBJ1: 90% accuracy on 50-100 page Swiss compliance documents
OBJ2: Zero-shot schema mapping with F1 > 85%
OBJ3: 40% hallucination reduction vs. baseline
OBJ4: Processing time reduced from 2-3 weeks to 1-2 hours
OBJ5: Validation at 3-5 Swiss financial institutions
OBJ6: Technology Readiness Level 7
OBJ7: 7-13B parameter models deployable on-premise
OBJ8: Validated on 500 multilingual documents (DE/FR/IT/EN)

Partners

Partner	Role	Contribution
FHGR	Research Partner	AI/ML research, model development (3,500h)
WeCanGroup SA	Implementation Partner	System integration, deployment (2,300h)

Project Timeline

2026                                    2027
Jan    Apr    Jul    Oct    Jan    Apr    Jul    Oct    Dec
|------|------|------|------|------|------|------|------|
[====== WP1: Project Management ========================]
[======= WP2: Domain Adaptation =======]
       [======= WP3: Document Understanding ======]
                     [======== WP4: Information Fusion ========]
                            [======== WP5: Document Pre-Filling ========]
  MS1    MS2         MS3         MS4         MS5

Milestones: - MS1 (M4): Project Foundation - MS2 (M6): Technical Validation - MS3 (M12): Mid-Project Review - MS4 (M16): Field Matching - MS5 (M20): Final Validation

Documentation

Project Wiki - Interactive documentation with PROJECT/PLAN/APPLICATION tabs
Progress Dashboard - Real-time task tracking
Project Goals - Quantifiable objectives and success metrics
Onboarding Guide - Start here if you are new to the project
Developer Setup - Environment setup and installation
Tools Reference - All scripts and tools documented
Glossary - Domain and project terminology

Repository Structure

wecan-innosuisse-ai/
├── Application/     # Source Innosuisse application (PDF)
├── data/            # Generated JSON data
├── docs/            # Project documentation
├── execution/       # Monthly task tracking (M01-M24)
├── kickoff/         # M1 kickoff materials
├── scripts/         # Core pipeline scripts
├── tools/           # GitHub integration utilities
└── web/             # HTML outputs (wiki, dashboard)

Quick Start

# Install dependencies
pip install pdfplumber

# Run full pipeline
python scripts/extract_pdf.py
python scripts/generate_project_plan.py
python scripts/update_wiki.py

# Open wiki
start web/wiki.html

Task Tracking

Progress is tracked via GitHub Issues with 48 consolidated issues (2 per month: FHGR + Wecan).

# Fetch latest progress
python tools/fetch_progress.py

# Generate dashboard
python tools/generate_progress_dashboard.py

# View dashboard
start web/progress.html

Funding Acknowledgment

This project is funded by Innosuisse - Swiss Innovation Agency under the Innovation Project program (Application 133.672 IP-SBM).

Contact

FHGR (Research): Prof. Joerg Osterrieder
WeCanGroup SA (Implementation): Vincent Pignon