<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Doric Stack Field Notes</title><description>Ideas into systems. Products, AI tooling, SaaS infrastructure, consulting, and technical writing from Doric Stack.</description><link>https://doricstack.com</link><item><title>BRICS payment rails are real, and they are at pilot scale</title><link>https://doricstack.com/field-notes/brics-mbridge-hype-vs-reality</link><guid isPermaLink="true">https://doricstack.com/field-notes/brics-mbridge-hype-vs-reality</guid><description>Five central banks are testing a settlement platform that does not touch SWIFT or the dollar. mBridge has been live since 2022, reached minimum viable product in mid-2024, and has moved about 55 billion dollars across roughly 4,000 transactions. SWIFT moves about 53 million messages on an average day. These are parallel rails for specific corridors, and infrastructure moves at infrastructure speed, not headline speed.</description><pubDate>Mon, 29 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Stablecoin payments have an architecture answer and no operations answer</title><link>https://doricstack.com/field-notes/stablecoin-exception-ops-gap</link><guid isPermaLink="true">https://doricstack.com/field-notes/stablecoin-exception-ops-gap</guid><description>At 3:17 AM the confirmation never arrived. Two million in institutional stablecoin, and the custody record said complete, the counterparty said pending, the smart contract said something else again. The regulatory stack is finally coherent. The exception playbook that ACH and wires spent decades building does not exist yet, and that gap is where institutions are walking in.</description><pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Sanctions screening as code: one policy, a different actuator per rail</title><link>https://doricstack.com/field-notes/sanctions-screening-as-code</link><guid isPermaLink="true">https://doricstack.com/field-notes/sanctions-screening-as-code</guid><description>The payment is final in under a second and your sanctions analyst is still opening the alert. On an instant rail there is no review window, so the screen has to clear or stop the payment in code, inside the latency budget. The policy model can be common across rails. The action on a hit cannot, because RTP, the card networks, and SEPA Instant each demand a different commit semantic.</description><pubDate>Mon, 15 Jun 2026 00:00:00 GMT</pubDate></item><item><title>The core was never the bottleneck. The operating model was</title><link>https://doricstack.com/field-notes/core-banking-operating-model</link><guid isPermaLink="true">https://doricstack.com/field-notes/core-banking-operating-model</guid><description>The plan said twelve months and the core replacement landed in year three. Nobody in that room was lying. They budgeted for a software install and got an operating-model change. After two decades on the payments, treasury, and integration side of these programs, here is the math I give clients before they sign a full legacy core replacement, and where the months actually go.</description><pubDate>Mon, 08 Jun 2026 00:00:00 GMT</pubDate></item><item><title>The broker pattern is older than the agentic-commerce headline</title><link>https://doricstack.com/field-notes/stripe-broker-pattern-older-than-headline</link><guid isPermaLink="true">https://doricstack.com/field-notes/stripe-broker-pattern-older-than-headline</guid><description>Stripe Shared Payment Tokens are the agentic-commerce headline, but anyone who ran a card tokenization rollout in 2014 recognizes the shape in five seconds. A scoped surrogate credential with a thin stable interface in front and brokered complexity behind is not new. Visa Token Service shipped it twelve years ago, and the architects who see the pattern early build the right systems instead of rebuilding every eighteen months.</description><pubDate>Mon, 01 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Context rot can start before the first prompt</title><link>https://doricstack.com/field-notes/mcp-broker-context-rot</link><guid isPermaLink="true">https://doricstack.com/field-notes/mcp-broker-context-rot</guid><description>My LLM session loaded 276,989 tool-schema tokens before I typed anything - eleven MCP servers, 609 tool definitions, 27.7 percent of a million-token window spent on payload the model might never call. A bigger context window did not fix that, it gave the problem more room. So I built mcp-broker, PgBouncer for MCP, and the always-loaded tool payload dropped by more than 80 percent.</description><pubDate>Wed, 27 May 2026 00:00:00 GMT</pubDate></item><item><title>BRICS does not need to kill SWIFT to change cross-border architecture</title><link>https://doricstack.com/field-notes/brics-local-currency-corridors</link><guid isPermaLink="true">https://doricstack.com/field-notes/brics-local-currency-corridors</guid><description>The interesting BRICS payments question is not geopolitics. It is architecture: when a US company pays a supplier in a BRICS country, does the payment still default to USD correspondent banking, or does the hub choose the corridor before release? A mature local-currency corridor adds a routing decision, and that decision is middleware, policy, liquidity, and evidence rather than a new button in the portal.</description><pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate></item><item><title>Under a second is the right answer to the wrong question</title><link>https://doricstack.com/field-notes/agentcore-payments-under-a-second</link><guid isPermaLink="true">https://doricstack.com/field-notes/agentcore-payments-under-a-second</guid><description>At an AWS Bedrock AgentCore Payments session the question was what happens at a thousand transactions a second, and the answer was settlement in under a second. Both are true and they answer different questions. The right frame for one payment crossing a trust boundary is the wrong frame for a million events a day inside one agent fleet, where the chain fee stacks up before anything else does.</description><pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate></item><item><title>The 11:47am question real-time rails ask of your ledger</title><link>https://doricstack.com/field-notes/intraday-position-event-log</link><guid isPermaLink="true">https://doricstack.com/field-notes/intraday-position-event-log</guid><description>What was the cash position at 11:47am? On a snapshot-based model the honest answer is an interpolation between two saved states. Real-time rails turned that guess into a liability, and the fix is to stop storing a position at all and fold an immutable event log into the answer. The event log is not free, and most teams have not put the bill on the budget.</description><pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate></item><item><title>Real-time payments break your treasury org chart before they break your tech</title><link>https://doricstack.com/field-notes/real-time-rails-break-treasury-org-chart</link><guid isPermaLink="true">https://doricstack.com/field-notes/real-time-rails-break-treasury-org-chart</guid><description>Your real-time rails settle in fifteen seconds. Your treasury org chart was built around an end-of-day cutoff. The gap between them is where the next reconciliation loss lives, and it shows up in three roles - cash positioning, exception ops, and intraday risk - long before it shows up in the tech stack.</description><pubDate>Mon, 11 May 2026 00:00:00 GMT</pubDate></item><item><title>Most AI agents never reach production. Payments is where that gets expensive</title><link>https://doricstack.com/field-notes/agentic-payments-production-gap</link><guid isPermaLink="true">https://doricstack.com/field-notes/agentic-payments-production-gap</guid><description>The widely cited figure is that 88 percent of AI proofs of concept never reach production. Agentic payments inherit that gap, and the teams that cross it are not winning on model selection. They are winning on payments-specific infrastructure that makes an agent idempotent, governable, and rollback-safe at the rail boundary.</description><pubDate>Mon, 04 May 2026 00:00:00 GMT</pubDate></item><item><title>Multi-rail is baseline now. The operational tax is the work</title><link>https://doricstack.com/field-notes/multi-rail-operational-tax</link><guid isPermaLink="true">https://doricstack.com/field-notes/multi-rail-operational-tax</guid><description>The multi-rail debate is over - most US banks enabling instant payments now run both RTP and FedNow. The cost moved from the strategy deck to the operations floor, where exception handling becomes a decision tree, reconciliation becomes real-time position-keeping across two liquidity models, and routing is a per-transaction decision in milliseconds.</description><pubDate>Tue, 28 Apr 2026 00:00:00 GMT</pubDate></item><item><title>ISO 20022&apos;s real deadline is a data migration, not a message change</title><link>https://doricstack.com/field-notes/iso-20022-structured-address-cliff</link><guid isPermaLink="true">https://doricstack.com/field-notes/iso-20022-structured-address-cliff</guid><description>The translator is ready, the mapping is tested, the structured fields are wired - and the project is still not on track, because the customer records are missing structured address data across KYC, core banking, and correspondent files. In November 2026, SWIFT, SEPA, and CHAPS start rejecting unstructured addresses, and most banks are framing the wrong problem.</description><pubDate>Mon, 20 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Payment modernization stalls at security review, not architecture</title><link>https://doricstack.com/field-notes/security-review-payment-modernization</link><guid isPermaLink="true">https://doricstack.com/field-notes/security-review-payment-modernization</guid><description>Payment modernization projects do not stall at the architecture layer. They stall at the security review layer - threat models built for monoliths, questionnaires written for general software, cycles tuned for annual releases. The teams that ship on time put security in the room before design, not after.</description><pubDate>Mon, 13 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Stablecoin custody and issuance converge on one charter</title><link>https://doricstack.com/field-notes/stablecoin-custody-issuance-charter</link><guid isPermaLink="true">https://doricstack.com/field-notes/stablecoin-custody-issuance-charter</guid><description>Two US regulatory threads just lined up. The GENIUS Act created a federal pathway to issue payment stablecoins, and the OCC&apos;s new rule lets national trust banks hold digital assets in non-fiduciary custody - then conditionally chartered Coinbase as one. Issuance and custody can now sit inside the same federally regulated counterparty.</description><pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Payment platforms have docs. Almost none have a developer experience</title><link>https://doricstack.com/field-notes/payment-developer-experience</link><guid isPermaLink="true">https://doricstack.com/field-notes/payment-developer-experience</guid><description>Payment integration usually starts with a PDF, a sandbox that takes days to provision, and a support email that answers in 48 hours - and the third email is where patience dies and a competitor gets Googled. I built a self-service workbench to close the distance between what a platform can do and what a developer can figure out alone.</description><pubDate>Mon, 30 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Multi-model is the new multi-rail</title><link>https://doricstack.com/field-notes/multi-model-llm-gateway</link><guid isPermaLink="true">https://doricstack.com/field-notes/multi-model-llm-gateway</guid><description>Every payment system routes across multiple rails - ACH fails, traffic moves to wire, and wire is dear for small amounts so you route to RTP. Most AI systems still call one model and hope it stays up. The LLM gateway is the same routing engine payments has run for decades, and it needs the same three things to count.</description><pubDate>Mon, 23 Mar 2026 00:00:00 GMT</pubDate></item><item><title>AI agents get two payment rails, card and stablecoin</title><link>https://doricstack.com/field-notes/ai-agent-payment-rails-split</link><guid isPermaLink="true">https://doricstack.com/field-notes/ai-agent-payment-rails-split</guid><description>Visa and Coinbase are building competing payment rails for machines, not people. Coinbase&apos;s x402 puts stablecoins on HTTP&apos;s long-reserved 402 status code for sub-cent agent payments. Visa keeps the card rails and their dispute resolution. The split is real, because a 10,000-call-an-hour agent and a $4,000 flight want opposite things.</description><pubDate>Mon, 16 Mar 2026 00:00:00 GMT</pubDate></item><item><title>A chartered bank just put a stablecoin on card settlement</title><link>https://doricstack.com/field-notes/stablecoin-card-settlement</link><guid isPermaLink="true">https://doricstack.com/field-notes/stablecoin-card-settlement</guid><description>SoFi Bank put SoFiUSD on Mastercard&apos;s settlement layer. Authorization still rides the card rails and the merchant sees nothing change, but the interbank fund movement happens on-chain, around the clock, with no batch window and no correspondent float. For the first time, issuers and acquirers get to choose their settlement medium.</description><pubDate>Mon, 09 Mar 2026 00:00:00 GMT</pubDate></item><item><title>The operating model kills payment modernization, not the tech</title><link>https://doricstack.com/field-notes/operating-model-payment-modernization</link><guid isPermaLink="true">https://doricstack.com/field-notes/operating-model-payment-modernization</guid><description>A bank buys a modern payment platform and keeps the 2019 team structure, the same escalation matrix, the same three-person on-call rotation. The platform ships daily. The governance model says 90 days. Most modernization failures are operating-model failures, and the org chart defeats the architecture every time.</description><pubDate>Mon, 02 Mar 2026 00:00:00 GMT</pubDate></item><item><title>A multi-rail exception mesh, not best-effort dedupe</title><link>https://doricstack.com/field-notes/multi-rail-exception-mesh</link><guid isPermaLink="true">https://doricstack.com/field-notes/multi-rail-exception-mesh</guid><description>A duplicate payment in production is not an edge case. It is the control plane showing you what you never modeled. A retried request can land as a final success on one rail and a settlement reject on another, look clean on the dashboard, and still move two accounts twice. Best-effort dedupe is not a control.</description><pubDate>Mon, 23 Feb 2026 00:00:00 GMT</pubDate></item><item><title>AI can suggest a payment, only the control plane completes it</title><link>https://doricstack.com/field-notes/ai-payment-runtime-control-plane</link><guid isPermaLink="true">https://doricstack.com/field-notes/ai-payment-runtime-control-plane</guid><description>A payment bot did not go rogue - the architecture did. A timeout produced two valid-looking instructions for the same payout, because a downstream retry never learned the first was already in flight. Mandate-based authorization like AP2 still needs a runtime control plane with real state discipline, not policy in a config file.</description><pubDate>Mon, 16 Feb 2026 00:00:00 GMT</pubDate></item><item><title>Spend limits are not delegated payment authority</title><link>https://doricstack.com/field-notes/delegated-payment-authority-agents</link><guid isPermaLink="true">https://doricstack.com/field-notes/delegated-payment-authority-agents</guid><description>An invoice agent hits a transient timeout, retries, and two payments clear to the same vendor. The model calls it success. Ops calls it a week of cleanup. Spend limits and roles are a speed bump. Real delegated authority is a delegation object, policy on every attempt, ledger-bound idempotency, and an instant kill switch.</description><pubDate>Mon, 09 Feb 2026 00:00:00 GMT</pubDate></item><item><title>Wire investigations become structured workflows</title><link>https://doricstack.com/field-notes/wire-investigations-structured</link><guid isPermaLink="true">https://doricstack.com/field-notes/wire-investigations-structured</guid><description>A wire investigation is the moment confidence breaks - the payment has moved and the clock is running. The Fedwire Funds Service changes planned for November 2026 turn investigation messages into structured workflows and add a Create Payment Return feature. Treat it as a rail feature, not a back-office patch.</description><pubDate>Mon, 02 Feb 2026 00:00:00 GMT</pubDate></item><item><title>A name check just became a priced control</title><link>https://doricstack.com/field-notes/payee-name-verification-priced</link><guid isPermaLink="true">https://doricstack.com/field-notes/payee-name-verification-priced</guid><description>The Fed&apos;s 2026 pricing notice introduces Payee Name Verification at two cents a transaction, starting on FedACH. It works by comparing a submitted payee name against names previously seen on that account. From an architect&apos;s view that is not a yes-or-no check - it is a new control point that reshapes routing, exceptions, and reconciliation.</description><pubDate>Tue, 20 Jan 2026 00:00:00 GMT</pubDate></item><item><title>The hard part of payments is access, not the API</title><link>https://doricstack.com/field-notes/payment-access-fed-rfi</link><guid isPermaLink="true">https://doricstack.com/field-notes/payment-access-fed-rfi</guid><description>A payroll rollout taught me the friction is never the integration. It is funding caps, routing permissions, and what can clear directly versus what has to detour through a sponsor bank. That is why the Fed&apos;s new payment-account RFI matters - access is becoming a design decision, not just a policy one.</description><pubDate>Mon, 12 Jan 2026 00:00:00 GMT</pubDate></item><item><title>API governance is payment architecture, not a footnote</title><link>https://doricstack.com/field-notes/api-governance-payment-rails</link><guid isPermaLink="true">https://doricstack.com/field-notes/api-governance-payment-rails</guid><description>A duplicate API call looks harmless until two payments clear and operations is left explaining the gap. The same error behaves differently on every rail - ACH hides it until batch, RTP and FedNow clear it instantly and final, wire forces a manual recall. On a multi-rail platform, governance is the architecture.</description><pubDate>Mon, 05 Jan 2026 00:00:00 GMT</pubDate></item><item><title>ACH&apos;s most boring 10 characters just became policy</title><link>https://doricstack.com/field-notes/ach-descriptors-payroll-purchase</link><guid isPermaLink="true">https://doricstack.com/field-notes/ach-descriptors-payroll-purchase</guid><description>From March 2026, Nacha standardizes the ACH Company Entry Description as PAYROLL or PURCHASE for those payment types. It sounds trivial, but a free-text field becoming a dependable category is what finally lets reconciliation automation stick on a rail that still moves tens of billions of payments a year.</description><pubDate>Mon, 29 Dec 2025 00:00:00 GMT</pubDate></item><item><title>Design open banking for the rule that keeps moving</title><link>https://doricstack.com/field-notes/cfpb-1033-open-banking-whiplash</link><guid isPermaLink="true">https://doricstack.com/field-notes/cfpb-1033-open-banking-whiplash</guid><description>The CFPB finalized US open banking under Section 1033 in October 2024, a federal court enjoined enforcement a year later, and the rule is now back in reconsideration. The data layer that survives that whiplash is the deliverable, not the API. Build for reversals, not launch day.</description><pubDate>Mon, 22 Dec 2025 00:00:00 GMT</pubDate></item><item><title>Euro payments now name-check the payee before they move</title><link>https://doricstack.com/field-notes/verification-of-payee</link><guid isPermaLink="true">https://doricstack.com/field-notes/verification-of-payee</guid><description>Since October 2025, euro credit transfers run a Verification of Payee check before they settle, and a name-IBAN mismatch triggers an alert. It hits US treasuries paying EUR vendors and payroll as hard as it hits EU banks, and the same name checks are a fair bet to reach FedNow and RTP next.</description><pubDate>Mon, 15 Dec 2025 00:00:00 GMT</pubDate></item><item><title>Instant rails now carry wire money. Do the controls?</title><link>https://doricstack.com/field-notes/instant-rails-wire-controls</link><guid isPermaLink="true">https://doricstack.com/field-notes/instant-rails-wire-controls</guid><description>FedNow matched RTP at a $10M network limit in November 2025 and added new risk rails - a receiver-account pre-check pilot, ScamClassifier reporting, an Exception Resolution Service, and 1,500+ banks live. Instant is treasury cash now, and batch-era controls just got a 24/7 bypass.</description><pubDate>Tue, 09 Dec 2025 00:00:00 GMT</pubDate></item><item><title>Agents do the wiring, architects carry the tradeoffs</title><link>https://doricstack.com/field-notes/ai-agents-payment-careers</link><guid isPermaLink="true">https://doricstack.com/field-notes/ai-agents-payment-careers</guid><description>AI agents are arriving faster than most banking career ladders can adapt, and the market is already rewarding people who work with them. The career moat was never wiring up another API call. It is the judgment calls agents cannot own - which rail to trust, how much latency to spend, who the regulator will be.</description><pubDate>Mon, 01 Dec 2025 00:00:00 GMT</pubDate></item><item><title>The payment API finance can still trust at quarter-end</title><link>https://doricstack.com/field-notes/erp-grade-payment-apis</link><guid isPermaLink="true">https://doricstack.com/field-notes/erp-grade-payment-apis</guid><description>The API worked. The books did not. A clean REST interface with pretty JSON still leaves accounting stitching exports for three days to answer one question - what happened to this payment across every rail and every failure path. ERP-grade APIs treat the contract as a shared language with finance.</description><pubDate>Wed, 26 Nov 2025 00:00:00 GMT</pubDate></item><item><title>AI agents can buy now, and fraud systems were not built for it</title><link>https://doricstack.com/field-notes/agentic-commerce-fraud</link><guid isPermaLink="true">https://doricstack.com/field-notes/agentic-commerce-fraud</guid><description>Stripe, OpenAI, and Google shipped protocols that let AI agents complete purchases inside a conversation. The rails are already fast enough. The problem is that fraud detection was architected around a human clicking buy on a trusted surface, and agents break that assumption.</description><pubDate>Tue, 18 Nov 2025 00:00:00 GMT</pubDate></item><item><title>ISO 20022 coexistence ends, and translation layers come due</title><link>https://doricstack.com/field-notes/iso-20022-coexistence-ending</link><guid isPermaLink="true">https://doricstack.com/field-notes/iso-20022-coexistence-ending</guid><description>Your bank says it is &quot;ISO 20022 ready&quot; for the November 22 cutover. Check the docs and it is usually a translation layer - MT to MX at the SWIFT boundary, MX back to MT inside. That is compliant, and it quietly throws away the structured data that was the whole point.</description><pubDate>Mon, 10 Nov 2025 00:00:00 GMT</pubDate></item><item><title>Variable recurring payments reset the subscription rail</title><link>https://doricstack.com/field-notes/variable-recurring-payments</link><guid isPermaLink="true">https://doricstack.com/field-notes/variable-recurring-payments</guid><description>Subscription billing has two bad options today - store the card or chase a manual approval every cycle. Variable recurring payments add a third: pre-authorized variable amounts with limits enforced by the rail. The UK is building it now, and the lesson for everyone else is about regulatory cycles, not technology.</description><pubDate>Mon, 03 Nov 2025 00:00:00 GMT</pubDate></item><item><title>Pay-by-bank&apos;s 3% savings hides a reconciliation bill</title><link>https://doricstack.com/field-notes/pay-by-bank-reconciliation-reality</link><guid isPermaLink="true">https://doricstack.com/field-notes/pay-by-bank-reconciliation-reality</guid><description>The CFO sees &quot;save 3 percent on transactions&quot; and the math looks obvious. Treasury sees continuous settlement at 3 AM and a reconciliation process built for batch. Before you switch to account-to-account, count the operating cost the savings slide leaves out.</description><pubDate>Mon, 27 Oct 2025 00:00:00 GMT</pubDate></item><item><title>DeepSeek-OCR reads documents as compressed vision tokens</title><link>https://doricstack.com/field-notes/deepseek-ocr-payment-documents</link><guid isPermaLink="true">https://doricstack.com/field-notes/deepseek-ocr-payment-documents</guid><description>Payments runs on documents - checks, wires, invoices, KYC files - and the OCR that reads them chokes on handwriting and damage. DeepSeek-OCR processes a page as roughly 100 visual tokens instead of a thousand text tokens, and it is open source. The headline accuracy comes with a compression ceiling worth knowing.</description><pubDate>Wed, 22 Oct 2025 00:00:00 GMT</pubDate></item><item><title>What actually breaks when you fine-tune a model</title><link>https://doricstack.com/field-notes/fine-tuning-failure-modes</link><guid isPermaLink="true">https://doricstack.com/field-notes/fine-tuning-failure-modes</guid><description>Train GPT-2 on Shakespeare and ask it the capital of France - it forgets Paris. That is catastrophic forgetting, one of several failure modes that surface the moment fine-tuning meets production, especially in regulated banking where &quot;the model flagged it&quot; is not an answer.</description><pubDate>Tue, 14 Oct 2025 00:00:00 GMT</pubDate></item><item><title>Fine-tuning is engineering with probabilistic outcomes</title><link>https://doricstack.com/field-notes/fine-tuning-reality-check</link><guid isPermaLink="true">https://doricstack.com/field-notes/fine-tuning-reality-check</guid><description>The vendor demo makes fine-tuning look like installing software - upload data, train, deploy. The real distribution of outcomes is brutal, most of the work is data preparation, and the model you trained for wire fraud can quietly forget how to read ACH. For financial services, guessing is not an option.</description><pubDate>Mon, 06 Oct 2025 00:00:00 GMT</pubDate></item><item><title>SpikingBrain bets on sparse firing over brute-force compute</title><link>https://doricstack.com/field-notes/spikingbrain-sparse-efficiency</link><guid isPermaLink="true">https://doricstack.com/field-notes/spikingbrain-sparse-efficiency</guid><description>Most large models keep every parameter lit for every token, like a city that never turns the lights off. SpikingBrain borrows the brain&apos;s trick - fire only the neurons that matter - and reports order-of-magnitude efficiency on Chinese MetaX chips. The headline numbers are real, with conditions worth reading.</description><pubDate>Thu, 18 Sep 2025 00:00:00 GMT</pubDate></item><item><title>Payroll virtual accounts break on the real-world edge cases</title><link>https://doricstack.com/field-notes/payroll-virtual-accounts</link><guid isPermaLink="true">https://doricstack.com/field-notes/payroll-virtual-accounts</guid><description>A clean virtual-account hierarchy looks simple until an HCM client asks you to reverse payroll for 200 employees across three states from three weeks ago. Multi-entity payroll is where account architecture meets 50 sets of wage law - and where instant rails do not yet fit.</description><pubDate>Tue, 16 Sep 2025 00:00:00 GMT</pubDate></item><item><title>Banks are starting to twin the org, not just the branch</title><link>https://doricstack.com/field-notes/organizational-digital-twins-banking</link><guid isPermaLink="true">https://doricstack.com/field-notes/organizational-digital-twins-banking</guid><description>Digital twins began as 3D models of physical space. The next step is simulating how a bank actually runs - so you can test a payment-rail change or a regulatory shift against every system before it touches production. Most banks still only document what exists.</description><pubDate>Mon, 08 Sep 2025 00:00:00 GMT</pubDate></item><item><title>The chip ban forced China to build smarter AI</title><link>https://doricstack.com/field-notes/chip-ban-algorithmic-efficiency</link><guid isPermaLink="true">https://doricstack.com/field-notes/chip-ban-algorithmic-efficiency</guid><description>The AI race everyone watches is the application layer. The one that decides it runs four layers down, in chip design, fabs, and export controls. When the US cut China off from the best GPUs, DeepSeek answered with efficiency instead of scale - and that is the lesson worth keeping.</description><pubDate>Tue, 26 Aug 2025 00:00:00 GMT</pubDate></item><item><title>Static payment routing could not handle the 2:47 AM wire</title><link>https://doricstack.com/field-notes/intelligent-payment-routing</link><guid isPermaLink="true">https://doricstack.com/field-notes/intelligent-payment-routing</guid><description>A $2.3M wire hit the system at 2:47 AM on a Friday, the rule saw &quot;large wire&quot; and chose Fedwire, and Fedwire does not run weekends. The transaction waited until Monday. The fix is not a faster rule - it is routing that understands business context.</description><pubDate>Mon, 18 Aug 2025 00:00:00 GMT</pubDate></item><item><title>Banks count fraud dollars and miss the exception bill</title><link>https://doricstack.com/field-notes/payment-exception-cost-center</link><guid isPermaLink="true">https://doricstack.com/field-notes/payment-exception-cost-center</guid><description>Every failed payment kicks off a chain of manual work that rarely shows up in a fraud dashboard. Across the top 15 US banks, dispute processing alone runs about $3 billion a year - and the wire flagged at $50K costs the same to investigate whether it is real or not.</description><pubDate>Wed, 13 Aug 2025 00:00:00 GMT</pubDate></item><item><title>GPT-5&apos;s unified model has a hidden token tax</title><link>https://doricstack.com/field-notes/the-hidden-ai-tax</link><guid isPermaLink="true">https://doricstack.com/field-notes/the-hidden-ai-tax</guid><description>GPT-5 ships as one model with an automatic router that decides how hard to think for you. Convenient, until you watch a simple question quietly trigger a web search, a page read, and a reasoning chain you never asked for - and burn through a capped Plus plan in an afternoon.</description><pubDate>Fri, 08 Aug 2025 00:00:00 GMT</pubDate></item><item><title>Consumer auth patterns fail the bank exam</title><link>https://doricstack.com/field-notes/ai-agent-security-architecture</link><guid isPermaLink="true">https://doricstack.com/field-notes/ai-agent-security-architecture</guid><description>Authentication that works for web scraping becomes a compliance problem the moment an AI agent moves real money. Enterprise payment systems need certificate identity, scoped service accounts, and short-lived tokens - the things a banking examiner expects to see.</description><pubDate>Thu, 07 Aug 2025 00:00:00 GMT</pubDate></item><item><title>Instant rails broke the treasury day</title><link>https://doricstack.com/field-notes/instant-rails-treasury-liquidity</link><guid isPermaLink="true">https://doricstack.com/field-notes/instant-rails-treasury-liquidity</guid><description>ACH cleared overnight, wires ran on business hours, and float was predictable. FedNow and RTP ended that. With 42 percent of RTP payments landing nights, weekends, and holidays, treasury teams are positioning cash by the hour - and the Fed discount window still keeps bankers&apos; hours.</description><pubDate>Mon, 04 Aug 2025 00:00:00 GMT</pubDate></item><item><title>Banks say AI catches the fraud. In production, the rules still do.</title><link>https://doricstack.com/field-notes/ai-fraud-detection-reality</link><guid isPermaLink="true">https://doricstack.com/field-notes/ai-fraud-detection-reality</guid><description>Nearly three-quarters of banks claim AI-powered fraud detection, but when a wire has to clear in under 100ms the model steps aside and deterministic rules make the call. The honest architecture is hybrid - and saying so is becoming a regulatory and legal requirement.</description><pubDate>Tue, 29 Jul 2025 00:00:00 GMT</pubDate></item><item><title>Banking&apos;s authentication crisis: AI defeated the voiceprint</title><link>https://doricstack.com/field-notes/banking-authentication-crisis</link><guid isPermaLink="true">https://doricstack.com/field-notes/banking-authentication-crisis</guid><description>Sam Altman told a Fed conference that AI has defeated most of the ways people authenticate today - voiceprints especially, clonable from a few seconds of audio. The fix is not a better voice system; it is rebuilding authentication around device-bound cryptographic keys.</description><pubDate>Wed, 23 Jul 2025 00:00:00 GMT</pubDate></item><item><title>Payment APIs are having their USB-C moment with MCP</title><link>https://doricstack.com/field-notes/mcp-payment-systems</link><guid isPermaLink="true">https://doricstack.com/field-notes/mcp-payment-systems</guid><description>Most payment integrations still look like 2015 - custom code per platform, months-long builds. The Model Context Protocol turns MxN integrations into M+N, and payment platforms are starting to ship it. The promise is real; so are the enterprise security gaps.</description><pubDate>Tue, 15 Jul 2025 00:00:00 GMT</pubDate></item><item><title>Most banking cloud migrations fail for the same architectural reason</title><link>https://doricstack.com/field-notes/banking-cloud-architecture-reality</link><guid isPermaLink="true">https://doricstack.com/field-notes/banking-cloud-architecture-reality</guid><description>Teams treat cloud like a faster data center instead of a different computing model. Banking workloads are not web apps - and lift-and-shifting a mainframe to the cloud buys you cloud costs with mainframe complexity.</description><pubDate>Mon, 14 Jul 2025 00:00:00 GMT</pubDate></item><item><title>APP fraud: real-time payments solved the wrong problem</title><link>https://doricstack.com/field-notes/app-fraud-speed-vs-safety</link><guid isPermaLink="true">https://doricstack.com/field-notes/app-fraud-speed-vs-safety</guid><description>Instant rails stripped out the friction that quietly prevented massive losses - and authorized push payment fraud walked straight through the gap. Once the money hits the rail it is gone, and multi-factor authentication cannot help when the victim authorizes the payment themselves.</description><pubDate>Mon, 07 Jul 2025 00:00:00 GMT</pubDate></item><item><title>Tokenized money is a 2025 architecture problem, not a 2030 one</title><link>https://doricstack.com/field-notes/tokenized-money-architecture</link><guid isPermaLink="true">https://doricstack.com/field-notes/tokenized-money-architecture</guid><description>The BIS just blueprinted a tokenized unified ledger and JPMorgan put a deposit token on a public chain - in the same week. The hard part for banks was never the blockchain. It is retrofitting core banking to carry programmable money without a big-bang migration.</description><pubDate>Wed, 02 Jul 2025 00:00:00 GMT</pubDate></item><item><title>The leadership red flags nobody talks about</title><link>https://doricstack.com/field-notes/leadership-red-flags-technical-teams</link><guid isPermaLink="true">https://doricstack.com/field-notes/leadership-red-flags-technical-teams</guid><description>Talented technical leaders derail their teams without realizing it. The warning signs are quiet - the team stops bringing bad news, the best people stop pushing back, and the resumes get updated before anyone says a word.</description><pubDate>Fri, 27 Jun 2025 00:00:00 GMT</pubDate></item><item><title>Cross-platform integration: the patterns that actually hold</title><link>https://doricstack.com/field-notes/integration-patterns-that-hold</link><guid isPermaLink="true">https://doricstack.com/field-notes/integration-patterns-that-hold</guid><description>Fiserv, Bottomline, and a core banking system each work fine alone. Wiring them into one payment chain is where projects go to die - and the difference between the ones that survive and the ones that become expensive lessons is a handful of boring resilience patterns.</description><pubDate>Wed, 25 Jun 2025 00:00:00 GMT</pubDate></item><item><title>Instant rails: the reality behind the marketing</title><link>https://doricstack.com/field-notes/instant-rails-reality</link><guid isPermaLink="true">https://doricstack.com/field-notes/instant-rails-reality</guid><description>RTP cleared $246 billion in 2024 and the headlines said instant payments had arrived. The average transaction sizes tell a quieter story - this is high-value institutional money on a faster rail, not the consumer revolution that was promised.</description><pubDate>Mon, 23 Jun 2025 00:00:00 GMT</pubDate></item><item><title>Your fraud AI can&apos;t see what your data hides</title><link>https://doricstack.com/field-notes/ai-fraud-fragmented-data</link><guid isPermaLink="true">https://doricstack.com/field-notes/ai-fraud-fragmented-data</guid><description>Fraud-prevention AI demos dazzle, then meet reality - the bank&apos;s wire, ACH, card, and instant-rail data live in separate systems. A model that cannot see the whole customer cannot catch a fraudster who moves across all of them.</description><pubDate>Thu, 19 Jun 2025 00:00:00 GMT</pubDate></item><item><title>RTP vs FedNow: the $10 million strategic divide</title><link>https://doricstack.com/field-notes/rtp-vs-fednow-strategic-battle</link><guid isPermaLink="true">https://doricstack.com/field-notes/rtp-vs-fednow-strategic-battle</guid><description>An $8M deal chopped into 16 payments because one bank used FedNow. With TCH-RTP at a $10M limit and FedNow at $500K, transaction size is quietly deciding which rail a corporate treasurer can use.</description><pubDate>Mon, 16 Jun 2025 00:00:00 GMT</pubDate></item><item><title>Why your payment modernization is backwards</title><link>https://doricstack.com/field-notes/payment-modernization-is-backwards</link><guid isPermaLink="true">https://doricstack.com/field-notes/payment-modernization-is-backwards</guid><description>Most banks modernize core-first and treat compliance as the finish line. With Fedwire going ISO 20022-native in July 2025, that order is exactly backwards - compliance is the foundation, not the finish.</description><pubDate>Tue, 10 Jun 2025 00:00:00 GMT</pubDate></item><item><title>Building a RAG pipeline that processes documents with incremental updates</title><link>https://doricstack.com/field-notes/rag-pipeline-incremental-document-updates</link><guid isPermaLink="true">https://doricstack.com/field-notes/rag-pipeline-incremental-document-updates</guid><description>A build note from a RAG document system tested on real content. The retrieval was the easy part; the document lifecycle - incremental add/delete and domain separation - was the real work.</description><pubDate>Thu, 05 Jun 2025 00:00:00 GMT</pubDate></item><item><title>SWIFT vs Ripple: Evolution, Not Revolution in Cross-Border Payments</title><link>https://doricstack.com/field-notes/swift-vs-ripple-evolution-not-revolution</link><guid isPermaLink="true">https://doricstack.com/field-notes/swift-vs-ripple-evolution-not-revolution</guid><description>The headlines say XRP and RLUSD will replace SWIFT. The likelier path is evolution, not revolution - a hybrid book where each rail earns the job it is actually good at.</description><pubDate>Wed, 04 Jun 2025 00:00:00 GMT</pubDate></item><item><title>The Hidden AI Revolution in Banking</title><link>https://doricstack.com/field-notes/hidden-ai-revolution-payment-routing</link><guid isPermaLink="true">https://doricstack.com/field-notes/hidden-ai-revolution-payment-routing</guid><description>The AI opportunity in banking is not another chat window. It is a decision layer that chooses the right payment path before money moves.</description><pubDate>Tue, 03 Jun 2025 00:00:00 GMT</pubDate></item></channel></rss>