Treasury questions are simple to ask. What will the bank balance be three weeks from now? Which of our bank accounts hasn't moved in months? Is days in receivables getting longer, and is that a customer problem or a billing problem? Are we paying vendors early for no reason, and are we taking the discounts we're offered? They're tedious to answer from NetSuite, because each one needs data from three or four places and a spreadsheet to hold the arithmetic.
The Sonar edition of my prompt library added five prompts to the Cash Flow and Treasury category, one for each of those questions and one that pulls the receivables side together into a plan. This post is a tour of what they do and what they insist on stating.
If you're new to this, Sonar AI is an AI agent that runs inside NetSuite. Every prompt in the library is a playbook that I engineered and tested against live NetSuite data, and you run it inside your own account, against your own records.
The Next Thirty Days
Predictive Cash Position projects the bank balance day by day for the next 30 days. It starts from the current balance, computed as the cumulative posting activity on bank accounts and corroborated against the balance sheet, adds the receivables due each day, and subtracts the payables due each day. Overdue receivables are disclosed separately rather than folded in, because money that was due last month isn't scheduled to arrive tomorrow.
The rule I'd point to is about collection probability. Not every invoice due on a given day gets paid that day, and the prompt won't pretend otherwise. Its default assumption is that 85% is collected on the due date and 15% slips one to five days, and it has to state that assumption in the report, because it isn't system data. Any projected day where the balance falls below a minimum threshold, $50,000 by default, is flagged. The output is a projection with its assumptions attached, which is the only kind of projection worth reading.
The Accounts Themselves
Bank Account Activity Profile looks at every bank account in the chart, active or idle, and profiles it on inflow, outflow, net, transaction count, current balance, and days since the last activity. Accounts silent for 60 days or more are flagged. A trailing six-month table shows inflow against outflow per account, with net-negative months highlighted. And it reports concentration: the share of inflow, outflow, and balance held in the top account, with a flag above 70%.
It's a modest prompt, and the findings tend to be modest too: a legacy account that should be closed, a subsidiary account that's carrying an idle balance, an operating account that everything runs through. But those are exactly the things a treasurer wants on one page, and NetSuite doesn't produce that page.
The Receivables Side
Days in AR Trend Monitor tracks days sales outstanding over time. It computes period-end receivables balances with a cumulative pattern rather than a snapshot, so the trend is built from consistent points, and it reconciles the aging buckets to the total before reporting any of them. Two conditions go to a person automatically: a DSO increase of more than five days in a single month, and receivables over 90 days exceeding 10% of the total. It won't invent an industry benchmark. If you don't provide one, the report says so and compares the business against its own history instead.
AR Collections Optimization and Working Capital Acceleration is the broader version. It diagnoses receivables health, works out why customers pay late, and produces a prioritized action plan with the working capital unlock quantified in dollars and a timeframe for each item. Every conclusion traces to a data lineage row. Every major conclusion carries a confidence level. And there's a list of things it will flag but never execute: customer escalations, account holds, credit limit or payment terms changes, write-offs, reserve recommendations, and any conclusion under 80% confidence on a material item. The report is written for a CFO who wants the answer first and the evidence attached, and that's the order it delivers them in.
The Payables Side
Vendor Payment Terms Optimizer looks at the other half of working capital. It compares stated terms against actual payment timing, vendor by vendor, and evaluates early payment discounts the only way they should be evaluated: as an annualized rate, compared against the cost of capital. A two percent discount for paying in ten days rather than thirty works out to roughly 37% a year, which is almost always worth taking, and the prompt does that arithmetic rather than reading the discount at face value.
Two of its rules are worth repeating to anyone who runs payables. Working capital freed by extending days payable is a one-time improvement, not an annual one, and the report has to say which it's quoting. And paying late isn't optimizing. Paying beyond stated terms without negotiating them damages relationships and eventually costs more than it saves, so the prompt recommends negotiation, not delay. Vendors paid 15 or more days early are flagged for review, because that's working capital leaking out for no return.
Wrapping Up
What ties the five together is the insistence on stating assumptions. The collection probability. The cost of capital. The benchmark, or the absence of one. The one-time versus annual distinction. Treasury work is full of numbers that look precise and rest on a guess, and the difference between a useful cash projection and a dangerous one is whether the guess is written down next to the number. These prompts write it down.
All five are in the paid tier of the Sonar AI Prompt Library, under Cash Flow and Treasury: Predictive Cash Position, Bank Account Activity Profile, Days in AR Trend Monitor, AR Collections Optimization and Working Capital Acceleration, and Vendor Payment Terms Optimizer.