The deployment of AI in business is probably one of the topics on every manager’s mind. Yet few actually share how they go about it. Last June, our CEO Christophe Saam spoke at a conference organised by INNOVAUD and AI Swiss on the theme “AI for SMEs: a practical guide”. There, he presented, in concrete terms, P&TS’s approach to deploying AI internally in a thoughtful and secure manner.
A Particularly Demanding Context
At P&TS, one of Switzerland’s leading advisory firms for innovation protection, every piece of data is by nature sensitive. Patents in the process of being filed, innovation protection strategies, information about inventions: the data we handle on a daily basis is subject to the strictest professional confidentiality.
Deploying AI without safeguards was therefore out of the question. Equally, staying away from such a transformative technology on the grounds of risk was not an option either. The objective lay elsewhere: to deploy AI in a genuinely useful way, delivering measurable added value to our teams, while never compromising the confidentiality of our clients.
Casparoff: A Purpose-Built AI Interface
To meet this objective, P&TS developed “Casparoff”: a unique, AI-powered web interface that provides access to all of the company’s data and allows it to be queried in natural language. A tool built from scratch within our own environment, for our team members.
An Agile Method, from MVP to Continuous Improvement
What sets the Casparoff project apart is as much the method as the result. Rather than launching a cumbersome technology project with an exhaustive set of requirements, P&TS chose a radically different approach:
- Fast start: A deliberately minimalist brief and a first MVP tested internally after just one week of development
- Dedicated talent: Timo Blattner, a recent university graduate, was specifically recruited to lead the development
- Bottom-up approach: Future users are involved at every stage of the product’s evolution
- Governance: A steering committee, including Charles-Edouard Bardyn and P&TS team members, sets priorities and validates strategic directions
The goal was not to replicate existing processes by injecting AI here and there, but to identify where AI can genuinely make a difference.
Data Security: Three Non-Negotiable Pillars
Data security was the guiding principle throughout the entire process. AI deployment in business, when done seriously, demands clear choices. P&TS built its approach around three fundamental pillars:
- Confidentiality: No training of models on our data, no permanent storage of prompts or responses outside our perimeter, strict limitation of transmitted information, and encryption of all access points.
- Data sovereignty and localisation: No data may be stored or processed outside Switzerland or the European Union. The jurisdiction of the Federal Data Protection and Information Commissioner (FDPIC) is contractually recognised by all partners involved.
- Technological independence: P&TS has not tied itself to any specific LLM, thereby preserving its freedom of choice and ability to adapt in the face of a rapidly evolving market.
These requirements were formalised according to a dual logic: by design, integrated from the outset into the interface’s architecture, and contractually with all project partners.
An Experience That Directly Benefits Our Clients
The Casparoff project is not merely an internal exercise: it directly feeds into the value P&TS brings to the companies it supports. This hands-on experience of AI deployment in business adds to an expertise that already makes a real difference: that of IT law and AI.
Pioneer Expertise: Where Law Meets Technology
What distinguishes P&TS in the legal support of AI projects is a dual mastery that few firms can claim. We do not simply master the regulations applicable to AI; we also understand the concrete technical and technological challenges our clients face.
This legal-technical approach allows us to go beyond purely legal or regulatory advice. We help businesses navigate applicable regulations, implement ethical practices for the development and deployment of their AI systems, and ensure transparency, fairness and accountability at every stage of their projects.
Comprehensive Support, from Strategy to Governance
In practice, our AI law advisory covers the full spectrum of business needs:
- Contracts and agreements: Drafting and negotiating all AI-related agreements, from technology licences to research partnerships, to protect your interests over the long term.
- Regulatory compliance: Ensuring your AI systems meet local and international legal requirements, and supporting you through your certification processes.
- Risk management and liability: Identifying the legal and ethical risks associated with your projects.
- Training and awareness: Training your teams (technical and managerial) on the legal, ethical and regulatory issues surrounding artificial intelligence.
- Best practices and audits: Helping develop AI solutions that meet the requirements of confidentiality, transparency, independence and sovereignty, with audits and recommendations to secure your implementations.
- Governance: Establishing usage rules, clarifying internal responsibilities, and providing legal support in the management of your AI development projects.
The Difference Practical Experience Makes
Having deployed AI internally (with the confidentiality requirements, technological trade-offs and regulatory constraints that this concretely entails) has allowed us to develop a unique expertise. This enables us to support our clients with genuine added value in deploying artificial intelligence, whether they are a start-up, an SME or a multinational.
