AI in treasury forecasting and cash positioning

Plain language meaning

AI in treasury forecasting and cash positioning helps bank treasury teams project cash balances, currency needs, settlement windows, nostro positions, collateral needs and funding actions across business lines and market conditions.

This topic is about bank treasury positioning and liquidity operations. It is not about corporate treasury dashboards only.

In a real bank, this use case is never just a clever model. It is a controlled banking capability. The bank must connect the source event, customer or account context, model input, AI output, operational action, compliance boundary, customer impact and retained evidence. AI and ML can improve detection, speed and consistency, but they do not remove the need for accountable decisions.

Where it sits in Applied Use Cases in Banking

This card belongs to Applied Use Cases in Banking. The working flow is Treasury source data, Cash position features, AI forecast, Treasury desk decision, and Position and funding evidence.

The correct way to study the use case is to ask what banking problem is being solved, what decision is influenced, who owns the outcome, what law or policy constrains the action and what record would satisfy risk, compliance, audit, operations and management review.

Banking data and evidence

The important data points are nostro balance, settlement calendar, currency exposure, expected payment flow, collateral position, funding rate, market event, and desk adjustment. These inputs matter because they can change customer treatment, operational queues, risk decisions, regulatory reporting, funding actions, dispute outcomes or investigation priorities.

The evidence pack should include position report, forecast output, assumption override, desk approval, funding ticket, limit status, and treasury MI pack. A strong bank can replay the use case from source data to model output, operational action, human review, final outcome and monitoring result. A weak bank only knows that AI suggested something.

Controls that make AI adoption safe

The core controls are source reconciliation, desk assumption review, limit monitoring, funding approval, stress overlay, manual adjustment log, and management reporting. These controls make the use case bank-grade because they tie the AI output to approved policy, source data, human authority, audit evidence, customer-impact controls and ongoing monitoring.

AI can assist by scoring risk, finding patterns, clustering events, summarising evidence, prioritising queues and suggesting next best operational action. It should not invent facts, clear regulatory alerts silently, make high-impact customer decisions without authority, weaken investigation quality or hide uncertainty behind a confident score.

Regulatory and governance lens

Applied banking AI must be read through model risk, operational risk, privacy, consumer protection, AML/CFT, sanctions, liquidity-risk management, accounting integrity, payment-system resilience and auditability. The relevant mix changes by use case, but the discipline is the same: the model supports a controlled banking workflow.

The practical test is simple. If a reviewer asks why the bank used the data, why the model output was trusted, why the customer received that treatment, why an alert was cleared, why a route was selected, why a forecast changed funding action or why a dispute was closed, the evidence must already exist.

Diagram walkthrough

Read the diagram from left to right as Treasury source data, Cash position features, AI forecast, Treasury desk decision, and Position and funding evidence. The diagram is a control map. It shows the minimum path by which a banking event becomes AI-supported insight, human or policy-controlled action and retained proof.

Use it as a 30-minute study method. For each box, ask what source system creates the data, what can go wrong, what control detects the weakness, who owns the action, what customer or regulatory impact could arise and what evidence proves closure.

Most important mistake to avoid

The common failure is making treasury AI look like a prediction tool, while bank treasury needs controlled positions, desk judgement, funding authority and evidence across currencies and settlement windows.

The correction is to keep the model inside the banking control structure. Speed is useful only when source lineage, decision authority, customer-impact review, audit trail, monitoring and issue ownership remain visible.

Source anchors for accurate study

FFIEC BSA/AML examination guidance describes suspicious activity monitoring as a risk-based process covering unusual activity identification, alert management, SAR decisioning, SAR filing and continuing-activity monitoring.

OFAC's Framework for Compliance Commitments describes sanctions compliance programme components including management commitment, risk assessment, internal controls, testing and auditing, and training.

Federal Reserve SR 26-2, dated 17 April 2026, gives revised model-risk guidance for traditional models and non-generative AI models used by banking organisations.

NIST AI RMF 1.0 uses Govern, Map, Measure and Manage functions, and NIST AI 600-1 adds risk actions for generative AI including source grounding, content provenance, security and human oversight.

Basel liquidity risk principles require banks to identify, measure, monitor and control liquidity risk and to project cash flows across assets, liabilities, off-balance-sheet items, currencies and stress scenarios.

CPMI cross-border payment work covers safety and efficiency of payment, clearing and settlement arrangements, ISO 20022 harmonisation, operating hours, payment-system access, interlinking and liquidity bridges.

CFPB supervision materials treat consumer complaints, actual consumer harm, fraud, disclosure compliance, information-security controls and supervised financial institutions as practical consumer-protection signals.

Regulation Z billing-error rules require defined credit-card dispute timing, investigation, consumer communication and treatment of disputed amounts while the error is unresolved.

ICC UCP 600 and related ICC guidance make documentary-credit processing document-driven and place strong emphasis on strict compliance, stipulated documents, refusal handling and banking practice.

Separating a forecast from a confirmed position

A treasury dashboard shows an opening balance, confirmed settlements and a forecast of later intraday flows. An AI model may estimate the timing of incoming receipts or outgoing obligations, but it should label those estimates separately from booked movements. The bank cannot use a predicted credit as if it were already available cash for settlement. Currency, account, value date and settlement system all matter to the position.

Suppose a large expected receipt is delayed. The forecast changes, but the confirmed balance remains what the account statement or settlement system reports. Treasury assesses the funding need and any approved transfer, while the model records its forecast version and the information available when it was made. After the window, the team measures timing error and amount error separately. An apparently accurate end-of-day total can conceal an intraday shortfall. The lesson should connect predictions to actual liquidity decisions, accounting records and exception handling without treating the model as a source of funds.

Banking practice note: customer purpose

For ai in treasury forecasting and cash positioning, customer purpose is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from nostro balance to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

AI can assist by ranking risk, finding weak signals, summarising case evidence, detecting behavioural shifts, grouping similar exceptions and preparing review notes. The bank should not allow a generated explanation, a confident score or a convenient dashboard to replace validation, human judgement, customer communication, regulatory decisioning or issue closure.

A strong implementation records the source event, data timestamp, permission or lawful basis, model version, feature values, score or generated output, threshold, reason code, user action, exception status, monitoring result, owner review and final outcome. That record lets risk, compliance, audit, technology, treasury and operations speak from the same facts.

Banking practice note: source lineage

For ai in treasury forecasting and cash positioning, source lineage is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from settlement calendar to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: KYC and account context

For ai in treasury forecasting and cash positioning, KYC and account context is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from currency exposure to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: transaction behaviour

For ai in treasury forecasting and cash positioning, transaction behaviour is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from expected payment flow to desk approval. Then ask which control from funding approval proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: model input quality

For ai in treasury forecasting and cash positioning, model input quality is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from collateral position to funding ticket. Then ask which control from stress overlay proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: threshold governance

For ai in treasury forecasting and cash positioning, threshold governance is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from funding rate to limit status. Then ask which control from manual adjustment log proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: human review

For ai in treasury forecasting and cash positioning, human review is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from market event to treasury MI pack. Then ask which control from management reporting proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: case management

For ai in treasury forecasting and cash positioning, case management is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from desk adjustment to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: customer impact

For ai in treasury forecasting and cash positioning, customer impact is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from nostro balance to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: regulatory reporting

For ai in treasury forecasting and cash positioning, regulatory reporting is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from settlement calendar to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: audit trail

For ai in treasury forecasting and cash positioning, audit trail is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from currency exposure to desk approval. Then ask which control from funding approval proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: privacy minimisation

For ai in treasury forecasting and cash positioning, privacy minimisation is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from expected payment flow to funding ticket. Then ask which control from stress overlay proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: false positives

For ai in treasury forecasting and cash positioning, false positives is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from collateral position to limit status. Then ask which control from manual adjustment log proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: false negatives

For ai in treasury forecasting and cash positioning, false negatives is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from funding rate to treasury MI pack. Then ask which control from management reporting proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: operational queueing

For ai in treasury forecasting and cash positioning, operational queueing is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from market event to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: management reporting

For ai in treasury forecasting and cash positioning, management reporting is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from desk adjustment to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: stress conditions

For ai in treasury forecasting and cash positioning, stress conditions is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from nostro balance to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: fallback operation

For ai in treasury forecasting and cash positioning, fallback operation is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from settlement calendar to desk approval. Then ask which control from funding approval proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: exception ownership

For ai in treasury forecasting and cash positioning, exception ownership is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from currency exposure to funding ticket. Then ask which control from stress overlay proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: root-cause analysis

For ai in treasury forecasting and cash positioning, root-cause analysis is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from expected payment flow to limit status. Then ask which control from manual adjustment log proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: model monitoring

For ai in treasury forecasting and cash positioning, model monitoring is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from collateral position to treasury MI pack. Then ask which control from management reporting proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: data drift

For ai in treasury forecasting and cash positioning, data drift is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from funding rate to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: policy control

For ai in treasury forecasting and cash positioning, policy control is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from market event to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: evidence retention

For ai in treasury forecasting and cash positioning, evidence retention is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from desk adjustment to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: reviewer authority

For ai in treasury forecasting and cash positioning, reviewer authority is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from nostro balance to desk approval. Then ask which control from funding approval proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: customer communication

For ai in treasury forecasting and cash positioning, customer communication is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from settlement calendar to funding ticket. Then ask which control from stress overlay proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: feedback loop

For ai in treasury forecasting and cash positioning, feedback loop is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from currency exposure to limit status. Then ask which control from manual adjustment log proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: risk appetite

For ai in treasury forecasting and cash positioning, risk appetite is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from expected payment flow to treasury MI pack. Then ask which control from management reporting proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: cost and service level

For ai in treasury forecasting and cash positioning, cost and service level is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from collateral position to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: control attestation

For ai in treasury forecasting and cash positioning, control attestation is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking workflow first and a model workflow second.

Trace one item from funding rate to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: customer purpose

Trace one item from market event to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: source lineage

Trace one item from desk adjustment to desk approval. Then ask which control from funding approval proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: KYC and account context

Trace one item from nostro balance to funding ticket. Then ask which control from stress overlay proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: transaction behaviour

Trace one item from settlement calendar to limit status. Then ask which control from manual adjustment log proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: model input quality

Trace one item from currency exposure to treasury MI pack. Then ask which control from management reporting proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: threshold governance

Trace one item from expected payment flow to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: human review

Trace one item from collateral position to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: case management

Trace one item from funding rate to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: customer impact

Trace one item from market event to desk approval. Then ask which control from funding approval proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: regulatory reporting

Trace one item from desk adjustment to funding ticket. Then ask which control from stress overlay proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: audit trail

Trace one item from nostro balance to limit status. Then ask which control from manual adjustment log proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: privacy minimisation

Trace one item from settlement calendar to treasury MI pack. Then ask which control from management reporting proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: false positives

Trace one item from currency exposure to position report. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: false negatives

Trace one item from expected payment flow to forecast output. Then ask which control from desk assumption review proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Banking practice note: operational queueing

Trace one item from collateral position to assumption override. Then ask which control from limit monitoring proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the use case is not yet bank-grade.

Separate prediction from a funding instruction

At a 09:00 treasury cutoff, forecast intraday net outflows by currency from scheduled payments, expected receipts and settlement windows. Keep actual ledger balance, committed obligations and predicted flows as separate fields. A model may predict that a large receipt is likely, but a funding decision should follow the bank's approved liquidity limits and human authority. Preserve the input snapshot and uncertainty when issuing the forecast.

Compare forecast errors across ordinary and stressed days, and by horizon and currency. A low average error can hide rare shortfalls on days when funding is most consequential. Introduce a delayed payment file and confirm that the service marks coverage incomplete and escalates. A later actual receipt does not change what the 09:00 forecaster knew. Record the forecast, final treasury action and realized cash separately for learning and audit.

Primary sources for further study

• BIS working paper: AI agents for payment-system cash management
• BIS research: AI and central-bank payment data