We audit the agent in the wild.
AI Risk Agentic & Diagnostics was built for one purpose: systematic pre-deployment radiography of autonomous AI agents — before hidden liabilities become capital losses.
Experts Bridging Law and AI
International lawyers, engineers, and risk managers trained at Oxford, Bologna, and Edinburgh are partnering with Geocapita Ltd in London, a risk management company for the insurance of your ecosystem of AI.
We use many technical tools specifically for any part of audits, like in cybersecurity, but it is important to mention the moral quality of the tools. For example, we use Oxetica, a platform created at the University of Oxford in partnership with the University of Bologna for the AI legal and ethical aspects in AI diagnostics, which means 9 centuries of tradition in legal frameworks and ethics, combined with the latest frontier level of research in Artificial Intelligence worldwide, which is a real guarantee for you and your company.
The key contact is:
Sofia Tedesco, an attorney at law (Juris Doctor), Master’s degree as an International Jurist from the University of Bologna (with an Executive MBA), and 15 years of experience in legal audits for embassies of 20 countries across 5 continents. She is a researcher in AI agents' innovation and autonomy at the Hai Lab & Anthropic 2026 of the University of Oxford, with studies in AI agentics, compliance, and regulation in AI at the business school of the same University.
Linkedin profile and contact :
linkedin.com/in/sofiatedesco
sofiatedesco1@yahoo.com
Key contact in Insurance for AI Agents :
www.climatecapital.tech
connect@climatecap.net
Trusted Globally
Built for live deployments, not theoretical models.
Most governance frameworks are written for models in a lab. We operate where the exposure is real — inside the 24/7 agent pipelines your operations already depend on.
Established in Canary Wharf and the City, we work inside the same regulatory and institutional environment as the clients we serve — financial services, insurance, and enterprise operations facing live-agent risk today.
The failure is the data.
Stress-tested under load, your agents reveal what no static audit can surface: hallucinations in production paths, compliance gaps at decision boundaries, and latent fraud vectors inside automated workflows.
Forensic visibility before your agents decide.
We deliver a structured pre-diagnostic — a radiography of your AI system — then scope the intervention your deployment actually requires. Not a framework. A finding.