- Essays··9 min read
The Fixed Cost That Nobody Priced
Anthropic's $518 billion compute obligation, disclosed in its September 2026 IPO prospectus, is not just a capital structure curiosity — it is the enterprise buyer's problem in miniature. Fixed minimum commitments are appearing in agentic platform contracts, and 93% of enterprises have already overshot their AI budgets before most have a production agent in service for six months. The finance owner who built the ROI case on per-task cost is holding a take-or-pay agreement nobody modelled.
agentic-aienterprise-financeai-infrastructuremodel-riskRead - Essays··13 min read
Sunday Essay — The Supervisor Who Missed the Branch Era
Three African central banks are opening AI supervision regimes, each with a different first sentence. Pretoria waits on the Financial Stability Board — Abuja has written AI into its anti-money-laundering code — Nairobi has published nothing while mobile lending runs at 1.5 billion shillings a day. The common gap sits outside all three perimeters.
ai-regulationfinancial-supervisionafricamobile-moneyRead - Essays··7 min read
The Actuary AIDA Never Meets
Halodoc's AIDA is a competent clinical assistant—the problem is structural. Every claim it touches flows to a national payer running a 108 per cent claims ratio, headed for payment default by July 2027, with no independent model-risk regime sitting between the model's output and the system's solvency exposure. Clinical AI in ASEAN scales in months; the actuarial and governance infrastructure that would hold it to account scales in years.
clinical-aimodel-riskindonesia-healthcarehealth-insuranceRead - Essays··13 min read
Sunday Essay — The Rate Cut, the Model, and Nubank's First Billion
Nubank cleared ten-figure net income for the first time in Q2 2026, with NuFormer — its financial-behaviour foundation model — now inside the underwriting decision, not merely supporting it. The retraining-cadence advantage over static scorecards is genuine. Whether the margin story holds depends on whether the Copom keeps cutting and whether delinquency migration stays within what the model expects — questions a 244 per cent coverage ratio answers for one bank, not for the market.
nubankbrazil-fintechai-credit-riskmodel-riskRead - Essays··8 min read
The Jakarta Claims Desk After the Copayment Rule
Indonesia's mandatory 10% copayment rule, in force since March 2026, has shifted the distribution of submitted health claims faster than any fraud-detection model calibrated on the old baseline can follow. The real discipline at a Jakarta claims desk is not the AI stack — it is who overrides it, whether those overrides are logged, and whether a coverage-adverse audit exists before the regulator requests one.
health-insuranceai-governanceclaims-processingindonesiaRead - Essays··11 min read
The agentic AI carve-out: why US and EU regulators are writing rules neither can enforce
The Federal Reserve placed agentic AI outside its model-risk framework in April; the European Commission issued transparency guidance two weeks before enforcement began in August. For banks operating across both jurisdictions, the collision is immediate and the governance gap is real. The first regulatory action will likely come for failing to document why deployment was safe at all, rather than for deploying these systems badly.
agentic-aifinancial-regulationmodel-riskeu-ai-actRead - Essays··12 min read
Sunday Essay · Banks Turned On the Contact-Centre Line Before the Standard Work Existed
Four US banks (Bank of America, JPMorgan Chase, Goldman Sachs and Wells Fargo) switched on production-grade generative-AI systems across their contact centres in the same quarter that federal supervisors carved these technologies out of revised model-risk guidance. Held against the Toyota Production System, the rollout reveals an industrial line running at scale with no jidoka cord to pull and no standard work to distinguish a good call from a bad one.
bankinggenerative-aimodel-riskai-regulationRead - Essays··9 min read
The model-risk carve-out: why US banks got a temporary reprieve from agentic AI oversight while EU lenders face their deadline
SR 26-2, the Federal Reserve's updated model-risk guidance, carves generative and agentic AI out of supervisory scope. The carve-out is a placeholder while regulators decide what replaces SR 11-7. The EU AI Act's high-risk obligations arrive on 2 August. For multinational banks, the result is a structural supervision paradox: US headquarters has time but no clarity; European subsidiaries have clarity but no time.
model-riskagentic-aieu-ai-actbanking-regulationRead - Essays··8 min read
Riyadh Re, the Cession Ladder, and the Actuarial Weight of a Guidance Note
Saudi Arabia's mandatory cession ladder reached 30% in January 2025; the UAE Central Bank's Guidance Note on responsible AI followed in February. Together they define what the AI story in Gulf insurance actually costs. A domestic reinsurance mandate sits on one side and a documented compliance overhead on the other: annual bias testing, Arabic-language disclosure, mandatory human review at the decision point. The vendor deck prices neither.
gulf-insurancereinsuranceai-governancesaudi-arabiaRead - Essays··11 min read
Sunday Essay · Singapore's Rulebook for Agentic Finance
Singapore's Monetary Authority has published the first regulator-authored document to name, at runtime, the exact control gates between an autonomous agent's proposed action and a live ledger. SAFR's policy engine, runtime validator and append-only audit ledger map onto what DBS and Grab Finance already run in production. For model risk officers from Jakarta to Frankfurt, this is the document to read before the vendor decks arrive.
agentic-aifinancial-regulationmodel-riskmas-singaporeRead - Essays··7 min read
Both regulators blinked · the gap sits exactly where the agents live
Two of the most consequential financial supervisors stepped back from AI oversight inside eight weeks. The retreat lands precisely on the generative and agentic systems that worry me most. The regulators blinked. The risk didn't even slow down.
model-riskai-regulationeu-ai-actagentic-aiRead - Essays··12 min read
Sunday Essay · Two Models, One Mistake: Notes on the Evaluator-Optimizer Pattern After PocketOS
The failure was never the bad command. It was the absence of a check between the command and the consequence. The evaluator-optimizer pattern is barely an idea, but it is the one still running on Monday morning.
evaluator-optimizerproduction-agentsmodel-riskagentic-evalsRead