Predictable Volatility is attending angelgroups on 17th September to explore practical questions for due diligence and early warning signs in board reporting.

Most AI failures aren't surprises. They're predictable, built from decisions that made sense at the time, encoded into systems that no one questioned, and validated by organisations that confused coherence for control. Predictable Volatility is a forensic intelligence practice for senior leaders in regulated industries, CDOs, CROs, board directors, and risk executives who need to understand not just what their automated systems do, but what they mean when something goes wrong. Our work draws on thirty years of practitioner experience across financial services, telecoms, and retail, and on the forensic study of failures that should never have happened: Post Office Horizon. Knight Capital. Boeing MCAS. The Global Financial Crisis. Wirecard. In every case, the warning signs were present. The architecture made them invisible. We work with leadership teams to identify where meaning has drifted, where human judgement has been quietly replaced by proxy, and where the gap between what a system was built to do and what it is now doing has become a governance liability. We offer the Meaning Audit (a half-day diagnostic for leadership teams), board-level advisory on AI decision risk, the Predictable Volatility book (May 2026), and a Substack series for practitioners. If your organisation runs automated decision systems and you haven't asked who owns the interpretation layer, this is where that conversation starts.

Founded 2024