"My personal bar for success: by the end of this project, a compliance officer spends more of their day on risk judgement, and less on document archaeology."More updates are coming, at a regular rhythm. Grateful to Vincent Pignon, Didier Baertschiger, Nathan Douet, Julia Jaussaud-Pignon, and the Wecan Group team for the partnership, and to Innosuisse for the trust.
The Team
Prof. Joerg Osterrieder
Principal Investigator
FHGR
AI and Finance research, extensive financial industry experience
Lennart Baals
Lead Developer
FHGR
Full-stack development and AI integration
Vincent Pignon
Founder & CEO
Wecan Group
Founded Wecan Group to bring digital compliance to Swiss finance
Didier Baertschiger
CTO
Wecan Group
Technical architecture and platform engineering
Nathan Douet
COO
Wecan Group
Operations and compliance product strategy
Julia Jaussaud-Pignon
CFO & CISO
Wecan Group
Financial governance and information security
Senior Scientific Advisory Board
Independent scholars advising the project on statistics, econometrics, digital finance, and AI methodology.
Prof. Wolfgang Karl Haerdle
Professor of Statistics
Humboldt-Universitaet zu Berlin
Research: high-dimensional time series, quantitative finance, blockchain.
"Rigorous statistics turns AI from an impressive demo into an auditable tool. A compliance Copilot has to show its work, and this project is built to do exactly that."
Prof. Codruta Mare
Professor of Statistics & Econometrics
Babes-Bolyai University, Cluj-Napoca
Research: spatial econometrics, FinTech, data science.
"Compliance teams do not need more data — they need models that survive the move from research into regulated practice. That is what this project sets out to deliver."
Prof. Daniel Traian Pele
Professor of Statistics & Econometrics
Bucharest University of Economic Studies
Research: financial markets, digital assets, time series.
"Onboarding and compliance in Swiss finance is a statistical-modelling problem wearing a paperwork costume. I am glad to support a project that treats it that way."
Prof. Maria Iannario
Full Professor of Statistics
University of Naples Federico II
Research: ordinal data models, FinTech, financial inclusion.
"Turning compliance from a box-ticking exercise into an evidence-based workflow is a worthwhile research goal — and a necessary one for the next generation of digital finance."
Dr. Hanna Kristín Skaftadóttir
Associate Professor and Program Leader of Business Intelligence
Bifröst University, Iceland
Research: AI and automation in business, business intelligence, digital finance, explainable AI, governance.
"A useful compliance Copilot must do more than extract data. It must support human judgment with reliable, verifiable, document-grounded outputs."
Prof. Monica Violeta Achim
Professor & doctoral supervisor in Finance
Babes-Bolyai University, Faculty of Economics and Business Administration, Cluj-Napoca
Research: financial analysis, economic and financial crime, money laundering, corporate governance.
"Anti-money-laundering and financial-crime prevention begin with the quality of the data captured at onboarding. A compliance Copilot that improves that first step is a meaningful contribution."
Prof. Claudia Tarantola
Full Professor of Statistics
University of Milan
Research: Bayesian methods, copula modeling, categorical data analysis, financial risk models.
"Bayesian and statistical-risk methods have matured far beyond the research lab. Applying them to the day-to-day work of Swiss compliance officers is where they earn their keep."
Prof. Alessandra Tanda
Associate Professor of Banking and Finance
University of Insubria, Varese
Research: FinTech and DeFi, ESG and climate risk in financial markets, governance of financial intermediaries.
"Obtaining the correct information for client onboarding is of the utmost importance when it comes to identifying customers and preventing money laundering activities. While using innovative tools to speed up the process is welcome, this must be done properly to ensure compliance and sustainable efficiency gains."
Prof. Stephen Chan
Associate Professor of Mathematics
American University of Sharjah
Research: extreme value analysis, distribution theory, digital assets, decentralized finance.
"Risk and compliance in a digital-assets era cannot be solved with spreadsheets. Bringing modern statistics and AI to the Swiss compliance stack is long overdue."
Prof. Roman Matkovskyy
Associate Professor in Finance and Accounting
Rennes School of Business, France
Research: cryptocurrency markets, systemic risk in digital assets, investor behaviour during market stress, Bitcoin and Ethereum pricing.
Dr. Yuanyuan Zhang
Assistant Professor, Mathematics and Statistics
American University of Sharjah
Research: cryptocurrency economics, decentralised finance, market efficiency, volatility, Bitcoin and stock-index pricing.
About the Project
Every time you open a bank account or apply for a financial product, your bank must verify who you are. This is called "compliance": the process of checking identities, reviewing documents, and ensuring everything meets legal requirements. Today, most of this work is done by hand: compliance officers read through stacks of passports, utility bills, and corporate documents, manually entering data into systems. It is slow, expensive, and error-prone.
That is the problem the Compliance Copilot sets out to solve. Funded by Innosuisse (the Swiss Innovation Agency), this 24-month research project (April 2026 through March 2028) brings together two partners: FHGR, the University of Applied Sciences of the Grisons in Chur, which contributes academic expertise in artificial intelligence, natural language processing, and financial technology; and Wecan Group, a Geneva-based fintech company whose digital compliance platform is already used by Swiss banks every day. The goal is to build an AI assistant that genuinely understands financial documents and can work alongside human compliance teams in a real production environment.
When a client submits their documents (a passport scan, a proof of address, a company registration), the Copilot reads and understands them automatically. It extracts the relevant information, checks it against regulatory requirements, and highlights anything that needs human attention. Because Switzerland has four national languages, the system is built from the ground up to handle German, French, Italian, and English documents with equal confidence. And unlike a generic chatbot, the Copilot is designed to flag uncertainty rather than guess. This is a critical requirement when wrong answers can have legal consequences.
Behind the assistant sits a focused research programme. The project advances the state of the art in document understanding for financial contexts, information extraction from complex multi-page dossiers, and, perhaps most importantly, techniques to reduce AI hallucination, the tendency of language models to produce plausible but incorrect outputs. In a regulated industry, every extracted data point must be traceable back to the source document, and the system must know when it does not know.
For bank customers, this means faster account opening, with days instead of weeks. For compliance officers, it means spending their expertise on judgment calls rather than data entry. For financial institutions, it means lower costs, fewer mistakes, and full data sovereignty, with all processing staying in Switzerland. And for the broader community, the project publishes its research openly: academic papers, tutorials, and datasets that advance the field beyond a single product.
The ambition is straightforward: turn compliance from a cost that financial institutions endure into an advantage they can build on.
Project Announcement
The Compliance Copilot project is a joint research and development initiative funded by Innosuisse, the Swiss Innovation Agency. The project runs for 24 months, from April 2026 through March 2028, and is carried out in partnership between FHGR (University of Applied Sciences of the Grisons) and Wecan Group SA. Its objective is to develop an AI-powered assistant that supports compliance officers at Swiss financial institutions in the processing of client onboarding documents.
The research partner, FHGR, is based in Chur, Switzerland, and contributes expertise in artificial intelligence, natural language processing, and applied financial technology research. FHGR is responsible for the core AI research, including domain adaptation of language models, document understanding, and hallucination reduction techniques. The implementation partner, Wecan Group, is a Geneva-based fintech company that operates a digital compliance and data management platform already used by Swiss banks and asset managers. Wecan Group is responsible for integrating the research outputs into a production-grade system embedded within existing compliance workflows.
The project team consists of six members across both partner organisations. On the FHGR side, Prof. Joerg Osterrieder serves as Principal Investigator and leads the research programme, and Lennart Baals serves as Lead Developer responsible for full-stack development and AI system integration. On the Wecan Group side, Vincent Pignon (Founder and CEO) provides strategic direction and industry relationships, Didier Baertschiger (CTO) leads the technical architecture and platform engineering, Nathan Douet (COO) manages operations and compliance product strategy, and Julia Jaussaud-Pignon (CFO and CISO) oversees financial governance and information security.
The project addresses four core research areas: first, domain adaptation of large language models for the Swiss financial and regulatory context, including training on multilingual documents in German, French, Italian, and English; second, document understanding, meaning the extraction of structured information from diverse compliance documents such as passports, proofs of address, commercial register extracts, and corporate filings; third, information fusion, which maps extracted data to the fields required by regulatory forms and internal banking systems; and fourth, hallucination reduction, developing techniques to ensure the AI system produces verifiable, source-grounded outputs rather than plausible but incorrect information.
For Swiss financial institutions, the Compliance Copilot is expected to reduce the time required for client onboarding from weeks to days, lower operational costs associated with manual document review, and decrease error rates in data entry. Compliance officers can focus their expertise on risk assessment and judgment-based decisions rather than routine data extraction. The system is designed to operate under Swiss data sovereignty requirements: all data processing takes place in Switzerland. For the Swiss financial sector more broadly, the project contributes to maintaining Switzerland’s position as a centre of regulatory innovation. All research outputs, including academic publications and methodological findings, are made available to the scientific community.
The project is funded through the Innosuisse Innovation Project programme, with a total budget exceeding CHF 500,000 covering research personnel, technical infrastructure, and development resources across both partner organisations.
Research
NLP, document AI, hallucination reduction, multilingual (DE/FR/IT/EN)
Industry
Swiss financial compliance, FINMA regulations, banks & asset managers
Timeline
24-month project, April 2026 – March 2028, 4 development phases
Team
6 members across FHGR (research) and Wecan Group (implementation)
Funding
Innosuisse (Swiss Innovation Agency), budget exceeding CHF 500,000
Impact
Faster onboarding, lower costs, Swiss data sovereignty, open research
Project Timeline & Cooperation
Apr – Sep 2026
Foundation & Domain Adaptation
Jul 2026 – Jan 2027
Document Understanding
Nov 2026 – Sep 2027
Information Fusion & Field Matching
Jul 2027 – Mar 2028
Integration & Validation
FHGR
Research Partner
University of Applied Sciences of the Grisons, Chur. Brings academic rigor in AI, natural language processing, and financial technology research.
Wecan Group
Implementation Partner
Geneva-based fintech company. Brings industry expertise in digital compliance and production systems for Swiss banks.
Together, FHGR provides the research foundation while Wecan Group delivers production-grade systems used daily by Swiss financial institutions.
Research
"The Compliance Copilot stands at a unique intersection, where decades of financial industry practice meet cutting-edge AI research. Having spent years working in the financial sector, I witnessed firsthand the enormous burden compliance places on institutions and their clients. As a researcher in artificial intelligence and finance, I now see an extraordinary opportunity: large language models, combined with domain-specific fine-tuning and rigorous hallucination controls, can fundamentally transform how compliance works. This project brings together the academic foundations of natural language processing, document understanding, and machine learning with the production-grade infrastructure of Swiss financial technology. It is not just an incremental improvement. It is a paradigm shift in how regulated institutions handle their most document-intensive processes."
— Prof. Joerg Osterrieder, Principal Investigator, FHGR
Wecan Group
Based in Geneva at the heart of Switzerland's financial industry, Wecan Group has spent years building the infrastructure that banks and asset managers rely on for digital compliance.
Their flagship product, WECAN Comply, is already used by Swiss financial institutions for managing client data, conducting compliance checks, and maintaining audit trails. The platform handles the full lifecycle of regulatory requirements, from initial client onboarding through periodic reviews.
Wecan's vision is "Compliance-as-a-Service," making professional-grade compliance tools accessible and efficient. Their team brings deep expertise in financial regulation, data security, and building production systems that meet the demanding standards of Swiss banking.
Production-Proven
WECAN Comply platform serving Swiss banks and asset managers daily
Security-First
Blockchain-based audit proofs and Swiss data sovereignty
Full Lifecycle
From client onboarding through continuous monitoring to audit preparation
Founded by Vincent Pignon, with a leadership team combining technology (CTO Didier Baertschiger), operations (COO Nathan Douet), and financial governance (CFO & CISO Julia Jaussaud-Pignon).
AI in Financial Compliance
A brief history of how technology has shaped regulatory compliance in financial services.
Pre-2000s
Paper Era
Compliance meant physical filing cabinets, manual document review, and handwritten checklists. Banks employed large teams to read every document by hand.
2000s–2010s
Digital Transition
Digitization brought electronic document storage and basic search, but the core work (reading, understanding, and verifying documents) remained manual.
2010–2018
Rule-Based Automation
Early automation used rigid rule engines: if-then checks against predefined patterns. Fast for structured data, helpless with unstructured documents like scanned passports or handwritten forms.
2018–2022
Machine Learning Wave
Computer vision and NLP models began handling OCR and basic entity extraction. Still required extensive training data and struggled with the diversity of real-world compliance documents.
2023–Present
LLM Revolution
Large language models changed the equation. Models that can read, understand context, and reason about documents make it possible to automate the judgment-heavy parts of compliance for the first time.
2026
The Compliance Copilot
Bringing together LLM capabilities with domain-specific financial knowledge and production-grade Swiss compliance infrastructure.
The timing is not accidental. Three forces are converging: AI models capable of understanding complex documents, mounting cost pressures on compliance departments, and regulatory frameworks that increasingly expect technological solutions. The Compliance Copilot sits at this intersection.
Why This Matters
The Compliance Copilot creates value across four dimensions.
For Academia
This project advances research at the intersection of natural language processing, document understanding, and financial regulation. It produces benchmark datasets, novel fine-tuning approaches for domain-specific LLMs, and published research that bridges AI theory and financial practice. The tutorials and open resources contribute to the broader research community.
For Industry
Compliance costs consume a significant share of operational budgets at financial institutions. The Compliance Copilot targets a 10x reduction in document processing time, faster client onboarding, and fewer manual errors. For banks and asset managers, this translates directly to lower costs and better client experience.
For Switzerland
Switzerland's financial sector operates under FINMA regulations that demand the highest standards. Swiss data sovereignty is non-negotiable: sensitive financial data must stay within Swiss jurisdiction. The Compliance Copilot is built with these requirements at its core, strengthening Switzerland's position as a leader in both financial services and responsible AI adoption.
For Citizens
When compliance works better, everyone benefits. Account opening becomes faster: days instead of weeks. Privacy protection improves through automated, auditable processes. And as financial institutions reduce compliance overhead, those savings flow through to better services and lower costs for customers.
Capabilities
AI-assisted client onboarding and document extraction
Streamline the client intake process with intelligent document analysis. Automatically extract and validate information from identity documents, financial statements, and compliance forms.
Continuous and periodic compliance checks
Move from manual spot-checks to continuous automated monitoring. Receive timely alerts when client profiles change or regulatory requirements evolve.
Intelligent analysis and secure collaboration
Leverage AI-driven insights to identify patterns and anomalies across your compliance workflows. Enable secure, auditable collaboration between compliance teams.
Audit preparation
Generate comprehensive audit trails and reports on demand. Reduce audit preparation time with pre-organized documentation and clear compliance evidence.
Tutorials
PhD-level tutorials covering the AI and engineering foundations behind the Compliance Copilot.
Latest Updates
News and announcements from the Compliance Copilot team.
JO
Two months in: why FHGR and Wecan Group are building a Compliance Copilot together.
A short follow-up to our February Innosuisse announcement, and a note on what I personally am doing on this project.
I am Joerg Osterrieder, Professor of AI and Finance at FHGR – University of Applied Sciences of the Grisons. For the last decade I have worked at the edge of data science and regulated markets, with colleagues across Switzerland and Europe, on the question of how machine learning actually earns its place inside a supervised institution. My role on the Compliance Copilot is the research side: the data, the evaluation discipline, and the parts that do not show up in a demo but decide whether a compliance team can rely on the output.
Wecan Group already operates compliance flows in production for Swiss and European banks, and that is precisely what research needs to stay useful rather than only publishable.
The problem we are going after is the compliance officer's day. Right now it is long PDFs, the same customer fields retyped between systems, and the same client record reconciled three times over. The Copilot is designed to take that mechanical layer off the desk so officers can spend their time on actual risk decisions.
CC
Established the Senior Scientific Advisory Board for the Compliance Copilot project, with nine international scholars in statistics, econometrics, digital finance, and AI methodology. Read more about the advisory board.
JO
Excited to share a milestone for our team at FHGR and Wecan Group: we have been selected by Innosuisse for a two-year project to build a Compliance Copilot, a dedicated AI agent designed to radically simplify compliance for Swiss financial institutions.
Our ambition is clear: transform compliance from a constraint into a strategic advantage. Together with Vincent Pignon, Didier Baertschiger, Nathan Douet, and Julia Jaussaud-Pignon at Wecan Group, we are building an AI-powered platform that supports the full compliance lifecycle, from client onboarding and document extraction, to continuous monitoring, intelligent analysis, secure collaboration, and audit preparation.
Innosuisse's support, both financial and strategic, with a total project budget exceeding CHF 500,000 is a strong validation of the vision we share with the Wecan Group team.
Initial demonstrations are already generating consistently positive feedback from early stakeholders.
Compliance is evolving, and the interface to compliance must evolve with it.