AIRL is a research lab first. Organisations work with it because of that research and the expertise behind it — not the other way around. Work happens in three forms, kept deliberately narrow.
Joint investigations, applied research and expert exchange with institutions, forensic teams and other researchers working on agentic AI risk. This is where an open question the lab is already pursuing meets an organisation with the operational access or incident data to help answer it.
Suited to research and forensic teams, academic and institutional partners, and specialist firms that need agentic AI investigation capability for a specific case.
Selected engagements where the lab's research — the methods and named frameworks set out on Methods — supports a concrete governance, resilience or risk decision: control architecture for a deployed agentic system, a forensic review of what a prior assessment missed, or embedded AI risk leadership during a defined compliance window under FINMA Guidance 08/2024 or the EU AI Act.
Each engagement is scoped to the specific gap, not sold as a fixed package. Regulated financial institutions, insurers and their counsel are the typical starting point.
Keynotes, panels, executive briefings and workshops on agentic AI control failure, behavioural governance, and the gap between compliance frameworks and how deployed systems actually behave. Sessions draw on the lab's primary research rather than generic AI-risk material. Delivered in English and German, in person and virtually.
Based in Zürich, working across Switzerland and the EU, in English and German.
Start a conversation →