About
A practitioner’s view, not a spectator’s.
I lead AI and agentic platforms at First Abu Dhabi Bank. This is the longer story behind The Philosophical Ledger — the work between an impressive demo and a system a regulator will sign off on, the beliefs that drive it, and the proof.
Abu DhabiThe Philosophical LedgerUpdated June 2026
Why the ledger? A ledger records judgments and their outcomes — what a regulator audits and a model learns from. That one idea runs through everything here. The essays are the human ledger: judgment captured from inside the work. The Situation Room is the machine ledger: signals captured, tested, and resolved into the few directions a leader must govern. Same thesis, two zoom levels — and governance is what lets either run.
01 · The story
How financial institutions decide — and why that became the work.
I lead AI and agentic platforms at First Abu Dhabi Bank. Before that, eleven years across Debt Capital Markets and fintech innovation at Bank of America taught me how financial institutions decide: slowly when trust is missing, quickly when architecture and accountability are clear.
That lesson is the whole thesis of this site. Intelligence is becoming infrastructure. The institutions that come out ahead will not be the ones with the best models — those are commoditising — but the ones that build the operating capacity to trust AI enough to act on it: governance, memory, evaluation, and approval woven into how the firm actually runs.
I build in the open because the argument should be inspectable. Agent OS and Ninja Harness turn traces, evaluation, memory, and approvals into working systems — not just polished answers. The writing here is the field record of doing that inside a regulated bank, not a commentary from the sidelines.
02 · The work
Between an impressive demo and a system a regulator will sign off on.
That gap is where I work: operating models, control planes, and institutional memory for AI that can act.
Operating models
Strategy, risk, technology, capital, and people moving together.
Control planes
Identity, authority, memory, observability, approval, and escalation.
Knowledge systems
Turning decisions and traces into intelligence the institution retains.
AI will expose weak institutions before it transforms strong ones.
03 · What I believe
Four convictions behind everything on this site.
Governance is the drivetrain, not the brake.
The firms that escape the amnesia tax treat governance as the thing that lets AI compound — capturing judgment, not just answers. It is not the layer that says no.
Institutions before intelligence.
Strong institutions compound AI; weak ones get exposed by it. The bottleneck is operating capacity, not model capability.
Traces, evaluation, memory, approval.
Trust is something you build and inspect, not something you declare. The system that can be tested, remembered, and signed off on is the one that gets to act.
Build in the open.
The argument should be inspectable, and so should the system. Open work is how a claim earns the right to be trusted inside an institution.
04 · Credentials & proof
The receipts.
- RoleSVP & Head of AI and Agentic Platforms, First Abu Dhabi Bank
- Experience18 years across investment banking, fintech innovation, and enterprise AI
- RecognitionGartner Innovation in Financial Services judge, 2024–2025
- ResearchPublished researcher on AI, agentic systems, and institutional trust
- Open sourceAgent OS & Ninja Harness on GitHub
- WritingThe Philosophical Ledger essay archive
05 · Follow the work
Get the Ledger.
One brief when the noise resolves into something worth acting on — verified signals, institutional implications, and no daily churn. Or reach out directly.