Free executive playbook.

Your deposits were never loyal to your bank — they were loyal to the effort it took to move them, and that effort is disappearing unevenly. Across the last rate cycle United States banks reported a 77% mean cumulative deposit beta on wholesale non-operational deposits from March 2022 to March 2024, measured on deposits of 7 days or less maturity (US Federal Reserve, March 2024 Senior Financial Officer Survey — a mean of self-reports from a panel of roughly 80 to 100 banks, not a market-wide measurement). The survey measures deposit betas and pricing rationale; it does not measure sweep-tool adoption, so why that beta ran so high is this playbook's inference rather than a survey finding: corporate treasuries running automated sweep tools extracted almost the entire rate rise, because nothing stood in the way. Silicon Valley Bank then showed the speed — $42 billion out in a single day, from a deposit base of which more than 94% was uninsured at the end of 2022 (Federal Reserve OIG Material Loss Review, September 2023). A networked, digitally reachable deposit base does not drain over 30 days; it drains inside a single day. But reachable is not the same as gone. A balance only leaves when three things line up: it is rate-sensitive, an agent can reach it, and the authorization gate lets it settle. Most banks measure the first two conditions and never map the gate — so they over-defend deposits that were never going to move and leave open the ones that already are. This playbook scores your book on the two axes, applies the gate, and names the segment an agent takes first — so you ring-fence that funding instead of defending a retail flood that has not arrived.

What's inside

  • The two wrong reactions, and the condition both skip — the panic error, which forgets that being visible to an agent is not the same as leaving, set against the complacency error, which trusts an inertia that has already collapsed for the balances that actually fund the bank; plus the three conditions a deposit needs before it can go — rate-sensitive, agent-reachable, and cleared by the authorization gate — and why missing any single one keeps the money where it is
  • The two axes, scored — A1 Rate-Sensitivity, the vertical axis: how hard a balance chases yield, its deposit beta, highest on surplus corporate and SME cash and rising on idle retail savings the moment a spread opens · A2 Agent-Reachability, the horizontal axis: whether an agent can see it and move it, a function of open-API and aggregator maturity, standardised in Vietnam by Circular 64/2024/TT-NHNN, whose own obligations commenced 1 March 2025 — the later date of 1 March 2027 closes a grandfathering window for connections that were already serving individual customers before that start, so it is a legacy-compliance deadline and not the moment the rails switch on — and constrained by its Article 7(3), which permits payment-initiation Open APIs to be opened only to banks and licensed payment intermediary service providers, so a consumer's own agent provider does not reach payment initiation without that licence
  • The control mechanic — F, the Loyalty Gate: reachable is not the same as gone. A deterministic authorization layer — biometrics, maker-checker approval, transaction limits and rate caps — decides whether a reachable, yield-hungry balance can actually settle out of your bank, which is why the two axes alone never predict flight
  • The Vulnerability Quadrant and the Monday ALCO drill — 4 segments scored on both axes and read through the gate — every high and low rating here is this framework's own analytic judgement, not a measured observation — from corporate and SME surplus cash (high sensitivity, high reachability, low friction: the high-risk quadrant today) through FX and uncapped balances and rising retail idle savings to retail operating cash held in place by primacy; then the 5-step drill — map the book, re-tier the surplus corporate and SME cash as rate-sensitive machine-fast funding rather than sticky cheap funding, move liquidity monitoring to intraday, become the agent yourself, and watch the gate
  • Where the gate is open, where it holds, and the dated clock — the 77% wholesale beta read as the preview of retail once its gate comes down; SVB's $42 billion single-day outflow (Federal Reserve OIG Material Loss Review, September 2023) against a 30-day survival ratio built for human speed; and what the Vietnamese moat actually rests on, where Circular 50/2024/TT-NHNN — as amended by Circular 77/2025/TT-NHNN, effective 1 March 2026 — requires OTP combined with biometric matching on transfers between different account holders above 10 million dong, or at or below that figure once the day's cumulative transfers pass 20 million, having absorbed the requirements formerly carried by Decision 2345/QĐ-NHNN, an instrument repealed in full by Decision 2872/QĐ-NHNN effective 1 January 2025 and therefore history rather than operative law; where Circular 48/2024/TT-NHNN's deposit-rate ceiling — whose numbers sit in Decision 2411/QĐ-NHNN, not in the circular itself — binds only demand deposits, tenors under 1 month and tenors of 1 to under 6 months, leaving deposits of 6 months and longer uncapped and market-determined, which is the band an automated sweep would target; and where Circular 64/2024/TT-NHNN's rails, PDPL 91/2025 and AI Law 134/2025 set the rest of the clock, with India's Account Aggregator framework — a consent-based financial-data-sharing rail, not a payment rail — crossing 100 million cumulative consents by August 2024, about 3 years after its 2021 launch, as a precedent for data reachability and not for an agent moving money

Five pages · free with a member sign-in · PDF.

Watch: youtu.be/-wzkVsYyjls

Want the weekly version? Subscribe to The AI Architect Letter · Read the latest insights.