Insights
Practical perspectives from senior work across banking transformation, architecture, AI, and new rails. Not commentary, observations from the decisions that matter.
Featured
Banks can add AI to every department and still leave the bank largely untouched. The bigger opportunity starts with redesigning the work, decisions and controls around the intelligence now available.
Read note →August 2026
6 min read
Banks can add AI to every department and still leave the bank largely untouched. The bigger opportunity starts with redesigning the work, decisions and controls around the intelligence now available.
Read note →September 2026
7 min read
Agents can reason, recommend and execute, but they cannot hold authority. The judgements only accountable humans may make have to be designed into the architecture, routed with evidence and recorded as state machines can act on.
Read note →March 2026
2 min read
Crypto, tokenisation, and new rails generate plenty of commentary. The harder question for banks is what they need to change in practice.
Read note →April 2026
3 min read
Stablecoins, tokenised deposits, CBDC-linked flows, and tokenised assets do not land in the bank the same way. If several matter, the bank cannot afford a separate middle for each one.
Read note →March 2026
2 min read
The question is not whether crypto matters in the abstract. It is where new rails solve a real banking problem strongly enough to justify the operating change.
Read note →April 2026
6 min read
Digital-asset strategy should follow the bank's footprint, product mix, and role choice. Copying a global-bank roadmap often imports complexity before there is a business case for it.
Read note →March 2026
4 min read
New rails do not just change how money moves. They change when liquidity becomes usable, how funding is timed, and what Treasury needs to control in real time.
Read note →March 2026
4 min read
A bank can keep digital channels open around the clock and still rely on delayed controls, overnight exception handling, and next-morning repair underneath. That gap matters quickly once new rails demand more continuous operating discipline.
Read note →March 2026
3 min read
Issuing a tokenised instrument is often the easy part. The harder question is whether the surrounding market infrastructure is strong enough to support it credibly.
Read note →March 2026
2 min read
A pilot should not only test whether the technology works. It should show whether the bank has a credible path to getting ready for what comes next.
Read note →March 2026
5 min read
The hard question is not whether a bank should build custody itself or use a partner. It is which control model the bank is prepared to own.
Read note →