Part I — The Foundation
Introduction
Every treasury team wants AI cash flow forecasting. Almost nobody is ready for it.
Imagine a mid-size manufacturer. Twelve subsidiaries across eight countries. Eight currencies. The CFO walks into the quarterly business review and asks: “Where will our cash be in thirteen weeks?”
What happens next is usually a spreadsheet nightmare. Treasury sends emails to every subsidiary requesting their forecast. Half respond on time. The data comes in different formats. Someone forgot to include a tax payment. Another subsidiary submitted numbers in the wrong currency. The team spends two weeks consolidating, cleaning, and reconciling — only to produce a forecast that is already outdated by the time it reaches the CFO’s desk.
This scenario plays out in thousands of companies every month. It has played out for decades. But something has changed. AI, and more specifically agentic AI, has matured to the point where it can genuinely transform how treasury teams forecast cash flows — as a practical tool that delivers measurable results.
At Automation Boutique, we have been building AI solutions for treasury and finance since the technology became mature enough to deliver real value. We have seen what works, what does not, and where most companies go wrong. The pattern is almost always the same: they jump straight to the AI part and skip everything that makes AI actually work.
AI cash flow forecasting is not magic. It is a system. A system that starts long before any model runs its first prediction. And if you get the foundation wrong, no amount of AI will save you.
This is the playbook. Step by step. From raw bank data to autonomous treasury operations. Whether you build it yourself, use spreadsheets, or use a platform like NineAnts, the principles are the same.
The business case: why cash flow forecasting matters
Cash flow forecasting is consistently ranked as the number one challenge in corporate treasury. Not payments. Not bank relationship management. Not FX hedging. Forecasting. Why? Because it is the foundation for almost every other treasury decision — and most companies are still doing it poorly.
Idle cash earns nothing. If you do not know that you will have €10M of surplus cash for the next sixty days, that money sits in a current account earning zero. At a conservative money market rate of 3%, that is €50,000 in missed yield in just sixty days. Multiplied across a year, the numbers become significant.
Emergency funding is expensive. When a cash shortfall catches you by surprise, you do not negotiate from strength. The spread between planned and emergency funding can easily be 200 basis points. On a €20M shortfall lasting sixty days, that is €65,000 in unnecessary interest — for a single event. With a reliable forecast, you can often smooth cash flows internally before going to the market at all: postponing non-critical payments, swapping liquidity across currencies, or using cash intercompany to cover short-term gaps.
Unhedged FX exposure hits the P&L. If your forecast does not accurately capture the timing and size of cross-currency cash flows, you cannot hedge effectively. A 5% adverse move on €10M of unhedged USD exposure is €500,000 directly off your bottom line.
And perhaps most damaging: poor forecasting means poor visibility into business health. Cash is the ultimate truth-teller. When customer receipts start declining or supplier payments start stretching, those are early signals. Without a reliable forecast, management discovers problems too late.
Companies with accurate, timely cash flow forecasts consistently achieve better investment yields, lower funding costs, tighter FX risk management, and earlier warning of business issues. Treasury stops being a reporting function and becomes a strategic partner. For most mid-to-large corporates, the annual value of improving forecast accuracy by even 20–30% runs into hundreds of thousands, sometimes millions of euros or dollars.
Look back before you go forward
Before you forecast a single dollar or euro, you need to understand your past. Historical transactions and bank account balances are the raw material of any cash flow forecast. Without them, you are guessing. With them, you are learning.
Aim for at least 24 months of historical transaction data. Why 24 months? Because seasonality — the recurring patterns that follow a calendar rhythm — only reveals itself when you have enough history to compare year over year. Tax payments hit in specific months. Customer receipts spike in Q4. Supplier payments cluster around quarter-ends. None of this is visible in three months of data. Thirty-six months is even better — more history makes it easier to distinguish real patterns from one-off anomalies.
For bank connectivity, the two main approaches are Host-to-Host (H2H) via SFTP (CAMT.053 or MT940 file drops onto a secure server, batch-based but reliable) and API connectivity (real-time or near-real-time, accelerated in Europe by PSD2 and Open Banking). For multinationals, SWIFT messaging adds another option for global banking partner networks. Most groups end up using a mix. Whatever the method, automate the pipeline and normalise the data into a single format. CSV uploads are a fallback, not a strategy.
Then categorise. Raw transactions are noise; categorised transactions are signal. Typical categories include Customer Receipts, Supplier Payments, Salaries, Taxes, Capex, Intercompany flows, Bank Fees, Interest, Rent, Utilities, Loan Drawdowns and Repayments, and Dividends. Make them specific to how your business actually operates — a SaaS company and a manufacturer have very different cash flow profiles.
Labelling every historical transaction is the hard part. Excel’s Power Query engine can handle rule-based categorisation for free; it works but requires a lot of rules and a lot of maintenance. The more scalable approach is hybrid: rule-based categorisation for obvious patterns, with AI classification picking up everything the rules miss. Watch out for intercompany transactions (a payment from your Dutch entity to your German entity can look like a regular supplier payment), refunds and reversals, and multi-currency transactions where the same supplier appears with different names depending on which bank is processing the payment.
Finally, choose your time buckets carefully. Daily liquidity needs a day-by-day view. Medium-term planning often works better weekly. Board reporting and long-range outlooks read better monthly or quarterly. A good forecasting system lets you switch between these views — store data at the most granular level (daily) and aggregate up. Once you flatten the data at source, you cannot go back.
Bring in what you already know
Your bank data tells you what happened. Your business systems tell you what is going to happen. Open invoices from your ERP, scheduled payroll runs, known tax obligations, lease payments, loan repayments — these are cash flows where the amount and timing are already largely determined.
But ERP data is not always as clean as it looks. Invoice due dates are not payment dates: your ERP says an invoice is due in 30 days but this customer historically pays in 45. A well-designed forecasting system learns this difference over time. Partial payments, credit notes, and disputes muddy the picture further. And different ERP systems (SAP, Oracle, Microsoft Dynamics) model payment terms differently, so consolidating across multiple ERPs requires normalisation.
Known cash flows form the baseline of your forecast. For the near term, they are usually quite accurate. But three months out, your known cash flows might cover only 30% of what will actually happen. That is where AI comes in.
Part II — The AI
Let AI fill the gaps
You now have two layers of data: categorised historical cash flows (what happened in the past) and known cash flows from connected systems (what you already know about the future). The gap between those known cash flows and reality is where AI earns its keep.
A well-designed AI agent analyses your historical data category by category to identify three types of patterns:
- Seasonality — repeating calendar-driven patterns. Customer receipts higher in November and December. Tax payments clustering in specific months. Supplier payments dipping every August when Europe is on holiday. Seasonality is one of the strongest signals because it is driven by structural factors that persist year after year.
- Trends — sustained directional movements. Customer receipts growing 10% year over year. Supplier payments gradually increasing as the business scales. A model that captures seasonality but misses the trend systematically underforecasts a growing business.
- Recurring patterns — regularities beyond seasonality and trends. Your largest customer paying on the 15th of every month like clockwork. Capex spiking every 18 months when equipment needs replacing. Dividends paid twice a year on fixed dates.
Understanding business context
Patterns alone are not enough. The system needs to understand business context. Take Rent. The company has been paying €50,000 per month for two years. Then suddenly, in March, the amount doubles to €100,000. A pure pattern-matching model treats this as an anomaly and tries to smooth it over. But the company signed a new lease for additional office space — €100,000 is the new normal. The model should not revert to €50,000 next month. It should project €100,000 going forward.
Now flip it: the same rent category jumps to €150,000 for one month because of a deposit on the new lease. That is genuinely a one-off and should not be baked into the recurring forecast. The difference between a structural change and a one-time event cannot be determined from numbers alone — it requires understanding what kind of cash flow you are looking at and what is driving the change. Rent behaves differently from Customer Receipts. Tax payments follow a regulatory calendar. Capex is lumpy and project-driven. Salaries are stable but step up when headcount changes. Each category has its own logic.
In NineAnts, the Treasury Agent classifies each category as stable, seasonal, trending, or volatile, and adapts its approach accordingly. When a pattern changes, it reasons about whether the change is structural or temporary. When the signal is genuinely ambiguous, it surfaces the change to the treasurer rather than guessing. That separates a forecasting tool from a forecasting agent: the tool applies a model, the agent thinks about the data.
Which AI models work best
Many forecasting systems apply the same model to every category. But Customer Receipts, with hundreds of transactions per month and clear seasonal patterns, needs a fundamentally different approach than Capex, which might have five transactions per year at irregular intervals.
Time-series models. Meta’s Prophet is particularly well-suited for cash flow forecasting because it was designed to handle multiple seasonalities (weekly, monthly, yearly), trend changepoints, and holiday effects — exactly the patterns that show up in business cash flows. ARIMA is a classic approach, effective for stationary data. Exponential Smoothing methods like Holt-Winters are simple but surprisingly effective for stable seasonal patterns, with the added bonus that they are interpretable.
Machine learning models. Gradient boosting (XGBoost, LightGBM) can incorporate external features like payment terms, customer segments, and macroeconomic indicators. Neural networks capture complex non-linear relationships but need substantially more data to train and are harder to interpret. For most treasury use cases today, time-series and statistical models still deliver the best balance of accuracy and interpretability — but that may shift as more data becomes available.
Simple statistical models. Don’t underestimate them. A well-tuned moving average or linear regression can outperform sophisticated models for categories with few transactions or very stable patterns. If your rent is €50,000 every month, you do not need Prophet. Matching model complexity to data complexity is engineering maturity, not compromise.
The agentic approach
Traditional forecasting trains a model offline on historical data, deploys it, and uses it until someone decides to retrain. The fundamental weakness: the model is always looking at stale assumptions. The world changes; the model stays frozen.
The agentic approach is different. The Treasury Agent does not train a model once and reuse it. Every time a forecast is requested, the agent analyses the data fresh — fetching the latest transactions, evaluating the characteristics of each category in real time, and selecting the most appropriate model on the fly, like a mechanic reaching for the right tool depending on the task. If a category that was stable for two years suddenly becomes volatile, the agent detects this on its next run and switches models. No waiting, no manual retraining, no deployment cycle.
In practice the agent picks: time-series models like Prophet for categories with strong seasonality and sufficient volume; recurring-pattern detection for stable, low-volatility categories like rent or insurance; schedule-based models for sparse, schedule-driven payments like quarterly taxes or annual dividends; and moving averages as the fall-back for categories with very limited data. The user can see which model was selected for each category and why. The model fits the data, not the other way around.
Human in the loop
Once the agent produces a forecast, a fundamental question emerges: how much control do you want to keep? “Human in the loop” is not just a safety feature — it is a philosophy about how humans and AI collaborate.
At one end of the spectrum, the AI is purely assistive. It generates predictions, but a human reviews and approves every entry. This is the right starting point for any organisation new to AI forecasting. In the middle, the AI operates semi-autonomously — filling in forecasts automatically for high-confidence categories, flagging uncertain predictions for human review. At the other end, the AI operates with high autonomy: handling routine categories end-to-end, escalating only when something unusual appears.
In NineAnts, autonomy is not a single low/high toggle. It is configured per action type: generating a forecast, writing a prediction directly into a category, adjusting existing entries, running scheduled jobs, sending notifications. For each, you decide whether the agent may act on its own, must ask for confirmation first, or is not allowed to perform the action at all. Most teams start in a review-first posture and unlock specific actions as confidence grows.
Regardless of autonomy level, every AI-generated entry is clearly marked. The user always knows which numbers came from the AI, which from connected systems, and which were entered manually. If you cannot distinguish AI-generated data from human-entered data, you cannot meaningfully review or trust the forecast.
From reactive assistant to proactive team member
Most AI tools are reactive. You open the application, type a question, get an answer. The AI waits. A truly agentic system can also be proactive: it can run on a schedule, generate the weekly forecast every Monday at 7am before treasury arrives, compare last week’s actuals to forecast every Friday afternoon and flag deviations, or monitor bank balances daily and alert the team when excess cash crosses a threshold. It does not wait to be asked. It does the work and reports back. This is the difference between an assistant and a team member.
The user experience
A cash flow forecast is only useful if the people working with it trust it. Trust comes from clarity. When a treasurer opens the forecast on Monday morning, they are looking at numbers that come from fundamentally different sources: ERP imports, bank statement data, manual entries from a colleague in Singapore, AI-generated predictions, and AI predictions that were manually adjusted by a local team. If all of these look the same on screen, the treasurer does not know what to question, what to verify, and what to trust.
Every number needs a label. Source type — AI, ERP, bank, manual, file upload — should be visible at a glance, and filterable. A full audit trail (every create, update, delete, import, AI injection, with timestamp and user) turns the forecast from a static spreadsheet into a living, traceable document. For many regulated industries, this is a compliance requirement.
Designing for collaboration across subsidiaries. When twenty subsidiaries contribute to a single consolidated forecast, the interface needs to serve both audiences without overwhelming either. Local users need simplicity: their categories, their numbers, what the AI suggests, approve or adjust. Group treasury needs depth: consolidated totals, drill-down by entity, filter by source type, and the ability to spot which subsidiaries have submitted and which have not. Central treasury should also be able to close a forecast cycle and lock it against further edits — coherence across entities is more important than squeezing in last-minute changes.
The role of the LLM in an agentic system
Traditional forecasting software is procedural: you configure it, press a button, it runs a model, gives you output. An agentic system is different. At its core sits a Large Language Model that acts as the brain. The LLM does not do the statistical forecasting itself — Prophet handles time-series modelling, moving averages handle sparse categories. The forecasting models are specialised tools. The LLM is the orchestrator. It decides which tool to use, when, and why.
When you ask the system to forecast Customer Receipts, the LLM first classifies your intent (forecast, analysis, comparison?), retrieves relevant data, examines the data characteristics, selects the appropriate forecasting model, interprets the model’s output in context, and communicates the result back in plain language — explaining not just the number but the reasoning behind it. The LLM can also reason about things statistical models cannot: whether a category is seasonal or volatile, whether recent data suggests a trend change, whether a discrepancy between forecast and actuals deserves attention. An LLM can read a transaction description, understand that “Office Lease Q2” is rent-related, and categorise it accordingly. A pure rule engine would need a specific rule for that.
AI security and data governance
Every treasury team considering AI should be asking hard questions about security — and should be sceptical of any vendor that brushes those questions aside. Financial data is among the most sensitive information a company holds: transaction histories, cash positions, bank account details, counterparty relationships, forecast assumptions. Mishandling creates real, board-level risk.
Where does my data go? You need to know exactly where your data is processed, not just stored. With AI, application data might sit in Frankfurt or Virginia, but if the AI model runs on servers in a different jurisdiction, your prompts and outputs may be crossing borders every time a forecast is generated. Ask for a clear statement that all processing — including AI inference — happens within infrastructure you can verify.
Is my data used to train AI models? Many providers use customer interactions to improve their models. Any serious vendor should give you an unambiguous contractual guarantee: your data is not used for model training, fine-tuning, or service improvement in ways that could expose it to others. There is a meaningful distinction between training a foundation model on your data (which should never happen) and using interaction signals like acceptance or rejection of a forecast to improve workflow logic for your own tenant. The second is how a system learns your preferences. The first is a red line.
Can other customers see my data? In a multi-tenant SaaS platform, the isolation question gets more complex when AI is involved. AI outputs are generated dynamically — if the model has been exposed to data from multiple tenants during processing, there is a risk of cross-contamination. Look for logical tenant separation at every layer including the AI layer, input validation and output filtering against prompt injection, and incident management procedures that specifically address AI-related risks.
Can I trace what the AI did and why? From a security and compliance angle, AI adds a specific audit requirement: you need to know not just that a number changed, but which AI model generated it and what data it was based on. Logging of AI-generated outputs should include the model or workflow used, the data sources referenced, and any subsequent human modifications.
How NineAnts handles this: all data is hosted on Microsoft Azure infrastructure, with data residency configured to match customer requirements (EEA-based data centres for European customers). AI processing, including all LLM interactions, runs within infrastructure that Automation Boutique controls. Customer data is never sent to third-party AI providers as independent recipients or controllers. It is not used to train, fine-tune, or improve AI models. Tenants are logically segregated at every layer. Automation Boutique maintains an ISO 27001 certified Information Security Management System, and our Data Processing Agreement includes a dedicated annex on AI security and processing controls.
The agent architecture
Think of it as a team. The LLM is the team lead. The forecasting models are the specialists. The data connectors are the researchers. The action engine is the analyst who translates findings into recommendations. The team lead coordinates them all, routes the right questions to the right specialist, synthesises the answers, and communicates with the user.
In practice, the system has dozens of specialised nodes, each handling a specific part of the workflow, with the LLM routing between them based on what the situation requires. Ask for a forecast, and you get one path. Ask why last month’s actuals deviated from the forecast, and you get a completely different path through the system. A treasurer might ask: “Why were our supplier payments so high in March?” That is not a forecast request. It is an analytical query that requires pulling actuals, comparing to historical norms, identifying outlier transactions, and forming an explanation. A procedural system cannot handle this. An agentic system can. That is the revolution in treasury technology — not just better models, better reasoning.
Part III — The Organisation
Multi-entity and multi-currency forecasting
Forecasting for a single entity in a single currency is one challenge. Consolidating forecasts across ten, twenty, or fifty entities in multiple currencies is a different game entirely. The complexity does not scale linearly — it multiplies. Three challenges in particular: eliminating intercompany flows, handling multiple currencies, and reconciling timing mismatches between entities.
Intercompany eliminations. Entity A in the Netherlands pays €500,000 to Entity B in Germany for management services. That is an outflow for A and an inflow for B — but at group level, it nets to zero. The cleanest way to handle this is to tag intercompany flows up front: every transaction between group entities goes into a dedicated Intercompany category. At a single-entity view they appear as normal flows; at group level the Intercompany category should net to zero. If it does not, that is your signal to drill down.
Multi-currency challenges. Each entity forecasts in its local currency. Group treasury needs the consolidated view in EUR or USD. Which exchange rate? Spot rates change daily — your three-month forecast moves even when nothing in the business has changed. Budget rates set at the start of the fiscal year give stability but diverge from reality. Forward rates represent the market’s expected exchange rate for a future date and are the most economically accurate for forecasting, though they require a forward-curve data source. For actuals, you can translate every transaction at the spot rate of the day it settled, or apply a single closing rate to the period — both valid, but they answer different questions about FX performance.
Timing mismatches. Entity A books a payment on Monday. Entity B receives it on Wednesday. In between, there is a two-day window where the cash has left one account but not arrived at the other. In the consolidated forecast, this creates a phantom cash gap that does not exist in reality. For intraday or daily forecasting, your system needs to understand that a payment and its corresponding receipt are the same transaction at different stages of settlement.
Centralised vs. decentralised forecasting
Bottom-up usually wins. Local teams have better visibility on local conditions. The most effective model is a hybrid: central treasury sets the framework (categories, time horizons, submission deadlines, quality standards) and local teams provide inputs within that framework. The AI validates everything.
AI as the quality layer
Decentralised forecasting introduces a quality problem. If twenty subsidiaries each submit forecasts, how do you ensure consistency? The Treasury Agent automatically reviews every submitted forecast against historical patterns and flags anomalies. A subsidiary in Brazil submits their quarterly forecast; the agent notices supplier payments are 40% below the same period last year with no corresponding change in customer receipts. It flags. The treasury team reaches out. The local team had forgotten to include a large vendor contract renewal due next month. These guardrails do more than catch errors — they educate users across the organisation about what a good forecast looks like.
Common pitfalls and how to avoid them
- Garbage in, garbage out. Spend the time on the foundation. Get 24 months of clean, categorised history before you turn on forecasting. It is not glamorous, but it is the highest-leverage thing you can do.
- Over-trusting the model. A forecast of €5M next week will usually be more accurate than €5M in one year. Always understand how confident the system is in its prediction, and treat different categories differently. Some forecast beautifully; others are inherently noisy.
- Ignoring business context. AI does not know about the acquisition you are planning next quarter, or the customer you are about to lose. The best forecasting systems make it easy for users to overlay their knowledge on top of the AI’s predictions. AI provides the baseline; human provides the context.
- Trying to forecast everything at once. Customer Receipts and Supplier Payments alone typically represent 60–70% of total cash flows. Start there. Add Salaries and Taxes next. Then the long tail. Prove the value, build trust, grow the scope deliberately.
- No feedback loop. If you never compare your forecast to what actually happened, you never improve — and worse, you never know how wrong you are. Close the loop. Every period, compare actuals to forecast. Measure accuracy by category. Identify systematic biases.
Part IV — The Value
From forecast to action
A cash flow forecast is not the finish line. It is the starting line. The entire purpose is to identify risks and opportunities early enough to act on them.
Excess cash. If your forecast shows surplus liquidity for the next three months, that cash should not sit in a zero-interest account. Money market funds, term deposits, or paying down a revolving credit facility — the right instrument depends on amount, duration, and your investment policy. An intelligent agent can analyse current money market rates across approved counterparties, check policy limits, and recommend a specific allocation. The treasurer reviews, approves, executes.
Cash shortages. If the forecast shows a funding gap in eight weeks, you have time to arrange a credit facility, draw on a revolving line, accelerate receivables, or delay payables. Eight weeks of lead time is the difference between a planned financing decision and a panic call to the bank.
FX exposure. The cash flow forecast reveals currency mismatches. If you are forecasting EUR expenses and USD receipts, you have FX risk. Identifying it in the forecast means you can hedge it before it becomes an FX loss on your P&L.
Early warning system. Most importantly, a cash flow forecast is an early indicator of business health. If a division’s cash flows are deteriorating — customer receipts dropping, supplier payments stretching — that is a signal. Management can act before it becomes a crisis. This is what turns treasury from a back-office function into a strategic partner.
The flywheel: actuals vs. forecast
AI cash flow forecasting is not a one-time exercise. It is a flywheel — and the flywheel is what separates companies that get marginal value from AI from those that get transformational value.
Every day, new actuals come in. They are automatically compared against what was forecasted. Where was the forecast accurate? Where did it miss? By how much? And why? This is variance analysis — traditionally one of the most time-consuming tasks in treasury. Most teams do it monthly, if at all. An AI agent does it continuously, detecting accuracy gaps and significant deviations: a large capital expenditure that was never forecasted; a customer receipt that came in two weeks late. It identifies patterns in the forecast errors themselves: are you consistently overforecasting supplier payments, underforecasting tax obligations?
These meta-patterns are surfaced to the treasurer as clear feedback to act on — adjusting assumptions, refining categories, or correcting source data so the next forecast cycle starts from a better place. Each cycle of forecast, actuals, analysis, and learning makes the next forecast better.
Forecast versions
A cash flow forecast is not a single static document. It evolves. Tracking versions matters for two reasons. First, accountability: when the CFO asks “how has our outlook changed since last month?”, you need to pull up both versions and show exactly what changed and why. Second, accuracy measurement: a forecast made six weeks ago should be less accurate than one made two weeks ago for the same period. If it is not, something is wrong with your process.
Cadence varies. Many teams version weekly — every Monday a fresh forecast incorporating the latest connected-system data. Others version monthly, aligned with the financial reporting cycle. Some create event-driven versions before a board meeting, a funding decision, or a major payment run. Each version is a snapshot — once finalised, it should be locked, so that when you look back at March 1st you see exactly what was forecasted on March 1st.
Forecast-vs-forecast analysis reveals something that actuals-vs-forecast cannot: the stability of your process itself. Are your forecasts converging as the period approaches? Healthy. Are they volatile right up until the last moment? Inputs are unreliable or your process is not capturing information early enough. It also reveals which categories are forecastable and which are not — and where to focus your improvement efforts.
Scenario analysis: what if?
A forecast tells you what is most likely to happen. Treasurers also need to understand what happens if things go differently. What if customer receipts drop by 20% next quarter? What if a major supplier changes payment terms from 30 to 60 days? What if EUR/USD moves 10% against you?
Traditional scenario analysis meant rebuilding three versions in Excel: best, worst, base. By the time you finished the worst case, the base case was already outdated. An agentic system makes this practical: you describe the scenario in plain language or adjust a few parameters, and the agent recalculates the entire forecast accordingly. The most valuable scenarios are compound: lose a major customer and the EUR weakens 5%; supplier payments accelerate and tax obligations come due. Each has cascading effects across the forecast — losing a customer affects receipts but also the FX exposure profile.
Part V — The Future
Where do we go from here?
Everything in this playbook — the data foundation, the connected sources, the AI forecasting, the multi-entity consolidation, the quality checks, the variance analysis — builds toward something bigger. If your forecast is accurate, and your agent identifies the optimal actions, the natural next question is: why not execute them?
The agent identifies €5M of excess cash for the next 45 days. It knows your investment policy. It knows current money market rates. It knows your counterparty limits. It recommends a specific term deposit with a specific bank at a specific rate. Today, a treasurer reviews and manually executes. The next frontier: within boundaries you set, the agent executes the investment, places the FX hedge, draws on the credit facility — and the treasurer’s role shifts from execution to strategy and exception handling.
Treasury becomes a real-time, continuously optimising function. Investment portfolios are continuously rebalanced as the forecast evolves. FX hedges adjust as exposures change. Funding is arranged before shortfalls materialise. Remember the CFO who asked “where will our cash be in thirteen weeks?” The answer used to take two weeks of spreadsheet consolidation and was already outdated when it arrived. Now imagine the same question. The agent already has the answer — updated this morning, broken down by entity, currency, and category, with a confidence range and a list of recommended actions attached.
That is the future this playbook is building toward. Not a single tool or model, but a system that turns financial data into decisions, and decisions into action — with humans setting the boundaries and AI doing the heavy lifting within them.