Audit. QC. Sandbox.

Provenance Labs

The safety layer for high-risk agentic AI. We help institutions and vendors prove that agentic systems are safe enough for the work that depends on them.

Start a conversation

The gap

Agentic AI is moving faster than its assurance layer.

Production AI systems increasingly rely on hierarchical multi-agent orchestration, not isolated agents. Safety assurance has to validate the full orchestration graph: which agents were invoked, what evidence they used, what decisions they made and whether each handoff complied with policy.

What we do

Clinical AI, where safety evidence has to stand up.

Provenance Labs works with NHS-facing clinical AI.

Post-market surveillance

Continuous safety and performance review as AI medical device obligations harden.

Clinical safety cases

Evidence structures aligned with clinical hazard logs and accountable deployment.

Orchestration traceability

Verification of agent pathways, evidence use, decisions and handoffs across high-risk workflows.

NHS procurement readiness

Audit outputs that support buyer confidence in high-risk clinical workflows.

Founders

Built by clinical and AI leaders.

Co-founder and CEO

Dr Ashley Merritt

Medical clinician and medical-AI specialist with more than 2,000 hours of hands-on experience across the pre- and post-training lifecycle for frontier AI labs, including writing, clinical review, quality assurance and team leadership.

A previous startup founder and entrepreneur, he has experience developing an AI triage application and working at the intersection of healthcare, technology and digital infrastructure. He has also represented a data-centre developer at Thailand's EEC Deep Dive Digital Forum, engaging hyperscalers, financial institutions and public-sector stakeholders.

Co-founder and CMO

Dr James Glasbey

Systems and digital health researcher focused on translating AI innovation into safe, useful clinical practice. Author of more than 250 publications and recipient of over £35 million in research funding, with extensive experience leading large multidisciplinary teams evaluating and implementing clinical AI models.

An experienced surgeon, founder and NHS innovator, he has helped deploy multiple AI applications within healthcare settings. His work includes collaboration with regulators and industry, including the MHRA AI Airlock team, alongside his role as healthcare domain lead for a medical AI training lab.

Contact

Talk to us about high-risk AI assurance.

We are speaking with clinical AI vendors, healthcare systems, frontier-model teams, regulators and partners working on accountable agent deployment.