Avtar Sehmbi, operating principles

Five restructures across four industries. Separating risk and control accountabilities at HSBC. Merging Centrica's trading, retail and wholesale models into shared services. Consolidating an investment bank onto a standards-led hub. Rationalising Deloitte's estate after Andersen. Building Global Capability Centres at Cigna.

Each was felt by people long before it was understood. Transformation holds when incentives, culture and accountability move together. Move one without the others and the change rarely lasts.

Consent orders and a Deferred Prosecution Agreement spanning global markets. Joint-venture shareholder groups across nine countries. Regulators on four continents. A technology estate consolidated onto shared services.

Four situations with little in common. Complexity is rarely reduced by understanding every detail better. Complexity is reduced by deciding what matters, saying so plainly, and staying with those priorities as the detail accumulates.

A Basel-aligned operating model built under supervision, separating first and second line accountabilities. A statutory board seat carrying risk and capital allocation. A business ready for the EU AI Act and DORA well ahead of enforcement.

Standards embedded into design rather than inspected for afterwards. Governance designed in accelerates delivery. Governance added later is harder to embed and slower to take effect.

Functions built from inception at Deloitte and HSBC, then scaled. Teams in the thousands across more than 60 markets. A statutory board seat overseeing executive succession and assessing whether leaders had built capability able to endure beyond them.

Give people work slightly beyond what they have proved, stay close enough to support the stretch, and be clear with them about where they stand. The legacy is the leaders you develop.

Chairing the Ethical AI and Cloud Centres of Excellence, governing adoption across more than 30 markets against clear measures of value. An API-first platform that earned its place by creating a new revenue stream. Machine learning that earned its place by finding threats siloed analysis could not see.

Every technology decision carried the same two questions. What changes for the business if it succeeds, and who owns it once the programme team leaves.