Treasury management proposals built from the prospect's own statements
A priced proposal in two days, not two weeks, every value cited to a statement line.
See one case, screen by screen ↓A treasury management proposal is what a bank gives a commercial prospect after reading its bank statements: the services the bank recommends, what each one is worth to the prospect, and what the package will cost each month after the earnings credit on its balances. Treasury Management Proposals produces one from the prospect's own statements — three cycles from any bank, in any shape — with every value cited to the statement line it came from, the price built at the bank's own list and tiers, and a named person approving the price before anything is sent.
AI use cases in banking: a buyer's guide to agentic AI for bank and credit union operations →
Updated
A prospect agrees to send three months of statements, and the work stops being banking and becomes clerical. The statements arrive as PDFs, as photographs of paper, or as another bank's account analysis statement with service codes down one column. The officer keys the volumes into a spreadsheet, guesses at what the prospect pays today, builds a pro forma by hand against a price list that lives in another system, works out an earnings credit on a calculator, emails a manager for a discount approval, and waits.
Ten to fifteen business days later the prospect receives a list of services. Nothing in it is tied to anything on their own statement, so nothing in it is arguable — and the commercial conversation it was meant to support has already moved on.
The price goes the same way. The discount is agreed in an email thread, so nobody can say afterwards what the officer's authority was, what the relationship justified, or when the price expires. And because nobody looked past the sale, implementation discovers the controls the proposal never considered: an exposure limit for the first ACH debit file, a suitability review for remote deposit, a security procedure that had to be agreed in writing.
Estimated split of the bank's working days on one mid-market treasury proposal, by hand and with the solution. The days the prospect spends finding the statements are not counted here — they are the same either way.
Seven agents read the statements, value each service and price it at the bank's list; an officer decides and a manager approves the price.
Calder Brook Veterinary Partners — 14 clinics, 420 staff, $47.2 million of revenue — is a prospect of Tessaly Community Bank, referred by the relationship manager who has a $6.4 million request in process. Three cycles of statements from their current bank arrive on October 14. Here is the afternoon of October 21, from the officer's screen. Every name and number is from the working solution's sample data.
The statement is on screen as a page, at the size you would read it, with their bank's name removed. Down the margin the agents have written five notes against the lines they came from: 612 paper checks a quarter, $1,142,380 of checks issued with no payee verification, 38 incoming wires at $15.00 each, $1,420,000 of collected balance earning $0.00 of credit, and card settlements arriving net of processor fees. Each note names the service that answers it and what it is worth a year.
Seven agents read 18 statement pages, grouped 2,689 items into four flows and found five things. One page could not be read and one document is missing — neither figure is in the total.
She clicks the checks figure. The source opens on what the agents read across all three cycles — June 198 items at $374,210, July 210 at $386,710, August 204 at $381,460; the quarter 612 items and $1,142,380, annualized to 2,448 checks and $4,569,520 — with the confidence and who opened it. Nothing in the proposal is a number she cannot put in front of a chief financial officer.
ACH payables saves $3,960 a year, and the page says exactly how: the Flows agent grouped the check detail into 318 payees and found 184 of them already accepting an electronic payment, which is 2,200 of the 2,448 annual checks, at $1.80 saved each — the published median cost of issuing a check less an ACH payment. Positive pay carries no saving at all, only the exposure it covers: $4,569,520 of checks issued a year. Moving the 152 incoming wires a year to $12.00 is worth $456, and the earnings credit on their own balances is worth $15,406 — $19,822 in all.
The card settlements arrive net of processor fees, so the effective rate cannot be read from the statement. The Recommendation agent's confidence fell to 0.62 against a 0.90 threshold, so it asked instead of answering: about $14,350 a year stays out of the total, the card processor statement is named as the thing it needs, and the request to the prospect is already drafted for her to send.
About $14,350 a year is excluded from the $19,822 until the document arrives.
Every page says what was read, at what confidence, and whether a figure in the proposal depends on it. July page 4 arrived as a photograph taken at an angle; it is listed as unreadable, no figure is taken from it, and the re-scan request naming the page and what it needs is written and waiting for her to send. The screen also states which privacy rules govern the pack she is holding.
Nineteen services at price list v14, each with its service code, volume, list price and proposed price: $1,145.00 a month against $1,329.41 at list. The discount is 13.87%, her authority is 10%, and the floor is $1,196.47 — so the pro forma says, before she can do anything else, that this is a pricing exception and who has to approve it.
$1,420,000 investable at 1.10% over a 30-day cycle is $1,283.84 — the formula is on the screen, not behind it. Against $1,145.00 of monthly service charges that is a net credit of $138.84: the whole package is covered by balances they already keep, and today they pay $812.50 a cycle for less, with no earnings credit at all.
First she declines one thing the agents raised: collecting the 9,400 wellness plans by ACH debit needs an exposure limit assessed before a file moves, so it is held for a second phase with reason DR-4 on the record. Then she sends the price for approval under PX-2, relationship value, with the $6.4 million request in process written where the manager will read it.
He sees what he is being asked for and what it buys: $19,822 a year found on the prospect's own statements, an earnings allowance of $1,283.84 a cycle, a net position of $138.84 in credit. He approves as priced, AP-1, with his own reason — and if he had drafted the proposal himself the system would have routed it one level up instead.
What we read, what we found and recommend, what it costs and what happens next — in plain words, with the arithmetic written out and nothing in it that is not from an accepted recommendation or the approved price. It could not have been sent before 2:48, and the record says so: approved by Garrett Okonkwo, AP-1, valid to December 20.
$19,822 a year found, a price a manager signed for, and a proposal sent seven days after the statements arrived.
The same proposal, seen by the four people who carry it — what their two weeks looked like, and what it looks like now.
Seven agents read the statements, value each service and price it at the bank's list; an officer decides and a manager approves the price.
Reads every statement page a prospect sends — a PDF, a scan, a photograph of paper, or an account analysis statement with service codes — and pins each figure to the page and line it came from. It removes the other bank's name from every page before anyone sees it.
Groups three cycles of activity into payables, receivables, balances and fees paid, with counts and values by cycle, and reconciles every group back to the statement totals to the cent. It is what turns a check column into 318 payees, 184 of whom already take an electronic payment.
Proposes services from the bank's own catalog, each with the value a year, the arithmetic, the published inputs it used and the citation back to the statement line. Where the document it needs is missing, it holds the value out of the total and names the document instead.
Builds the pro forma at the bank's list prices and volume tiers, computes the earnings credit rate allowance from the prospect's own balances, measures the discount against list, and tests it against the authority of the person who drafted it.
Writes the proposal in plain words from the services the officer accepted and the price that was approved: what we read, what we found, what it costs, what it is worth, what happens next — with the citations attached.
Names what must be true before the first file moves — the ACH exposure limit, remote deposit suitability, beneficial ownership on file, the agreement and its product schedules, entitlements and a test file — with the rule and date behind each one.
Watches the pipeline: statements outstanding, pricing approvals against the one-day clock, proposals opened but unanswered, and prices about to expire. It drafts the reminder, naming what is outstanding.
Approves every price and every proposal sent; never his own. The agents propose; a named person decides.
Ask about anything on the statement, the recommendation, the pro forma or the queue — or say what to prepare. Every answer carries the lines it came from, and then does the thing.
Why does ACH payables save $3,960 a year?
Their statements show 612 checks paid in the quarter — 2,448 a year, $1,142,380 last quarter. The check detail groups into 318 payees, and 184 of them already receive an ACH credit from someone, which means they accept an electronic payment. Those payees account for 2,200 of the 2,448 annual checks. The saving a payment is $1.80: the published median cost of issuing a check is $2.01 to $4.00, against $0.26 to $0.50 for an ACH payment. 2,200 × $1.80 = $3,960. The other 248 checks a year go to payees with no electronic record, so they stay on paper — which is why positive pay is recommended beside it.
Can I price ACH credits at seventeen cents?
Not on your own. List is $0.22 on price list v14, and at $0.17 the package comes to $1,145.00 against $1,329.41 at list — 13.87% below. Your authority is 10%, a floor of $1,196.47, so this is a pricing exception and the sales manager approves it. The reason that will carry it is relationship value: the $6.4 million request in process, and an earnings allowance of $1,283.84 a cycle that more than covers the $1,145.00.
Which of the proposals waiting is worth its discount?
Four wait for a price before Calder Brook joins them. Westvale Orthodontics at 13.38% and Calloway Bend Logistics at 15.94% are both past the drafting officer's authority, so each needs a reason the sales manager can read. Harrowgate Auto Group at 14.6% he drafted himself, so it has routed one level up. Only Greystone Ridge Builders at 3.82% is inside authority — the one the bulk action will take. Thirteen of the 34 open proposals carry a discount above 12%, which is what the November price review should read first.
What is missing from this pack, and what does it cost us?
Two things. July page 4 arrived as a photograph at confidence 0.42, below the 0.90 threshold, so no figure in the proposal comes from it and the re-scan request is drafted. And the card processor statement is not in the pack: their card settlements arrive net of processor fees, so the effective rate cannot be read. That leaves about $14,350 a year out of the $19,822 — a number that cannot be cited does not go in a proposal. One month of the processor statement is enough.
13 screens from the working solution, on its sample data. Pick one to see it large.
Nothing about the pricing is ours. The services, the list prices, the volume tiers, the earnings credit tiers and the discount authority are settings with an owner and a version, and every change is saved with a reason.
| Setting | Default | Choose from |
|---|---|---|
| Services and list prices | 42 services with their service codes, version 14 | Read-only here, owned and versioned by the treasury product manager |
| Volume tiers | Three, by monthly items | Three tiers, four tiers, or none |
| Earnings credit tiers | 0.85% · 1.10% · 1.25% by balance | Read-only here, owned and versioned with the price list |
| Officer discount authority | 10% of list | 5%, 10% or 15% of list; the manager and the level above follow the same matrix |
| Maker-checker on every proposal sent | On | Cannot be switched off |
| Agent confidence threshold | 0.90 | 0.85, 0.90 or 0.95 — below it an agent asks instead of answering |
| Proposal validity | 60 days from approval | 30, 60 or 90 days |
| Prospect statements kept | 180 days if the proposal is not accepted | 90 days, 180 days or a year; five years from the end of the relationship if it is accepted |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “estimated” = our arithmetic from published inputs, with the working shown on the screen · “target” = the design goal, measured in the pilot · “typical” = a published figure, cited. Sources: AFP 2022 Payments Cost Benchmarking Survey, reported by Nacha — median cost of issuing a paper check $2.01–$4.00 against $0.26–$0.50 for an ACH payment · AFP 2026 Payments Fraud and Control Survey — 58% of organizations reported check fraud in 2025 · McKinsey & Company, Turbocharging revenue growth in US treasury management — $15–20 million of growth potential for an average regional bank · Nacha Operating Rules, originator assessment and exposure limits, with the fraud-monitoring rules effective March 20 and June 19, 2026 · FFIEC Risk Management of Remote Deposit Capture, January 14, 2009 · FinCEN customer due diligence rule, 31 CFR 1010.230 · Regulation P, 12 CFR 1016, and the GLBA section 501(b) interagency information security guidelines. Every prospect, person, clinic, payee, account and figure on this page is invented sample data; the prospect's current bank is never named.
A treasury management proposal is what a bank gives a commercial prospect after reading its bank statements: the services the bank recommends, what each one is worth to the prospect, and what the package will cost each month after the earnings credit on its balances. Treasury Management Proposals produces one from the prospect's own statements — three cycles from any bank, in any shape — with every value cited to the statement line it came from, the price built at the bank's own list and tiers, and a named person approving the price before anything is sent.
The agents read every page the prospect sends, pin each figure to the page and line it came from, and group three cycles into payables, receivables, balances and fees paid, reconciled to each cycle's own totals. That turns a column of checks into 318 payees and the 184 of them who already take an electronic payment, and turns an incoming-wire fee line into what the same wires would cost at your prices. A page that cannot be read is listed as unreadable with a re-scan request, and no figure is taken from it.
On the prospect's own balances, with the arithmetic on the screen: average ledger balance less average float gives the investable balance, multiplied by the earnings credit rate from your tiers, by the days in the cycle, divided by 365. In the working solution $1,420,000 at 1.10% over a 30-day cycle is an allowance of $1,283.84 against $1,145.00 of monthly service charges — a net credit of $138.84, which is the strongest line in the proposal.
Every service at your list price with its service code and the volume from the statements, the volume tier applied, and the proposed price beside the list price. The discount is measured against list and tested against the drafting officer's authority. Inside it, the manager still approves the proposal before it is sent; beyond it, the exception routes to the manager with a reason code and the relationship context attached — and a proposal the manager drafted routes one level up instead.
They are held under your information security program and the consent the prospect gave when it sent them, and kept for 180 days if the proposal is not accepted, or five years from the end of the relationship if it is. Both are settings. Where the prospect is a sole proprietor rather than a company, the pack is treated as consumer information and the consumer retention and redaction settings apply — and the screen says which rule is in force.
Neither. The account analysis system stays the billing engine and the source of your service catalog and codes; the relationship pricing system stays the source of list prices, tiers and the authority matrix; the digital banking platform still runs the services once they are live. This sits in front of all of them, at the point where a prospect's statements become a priced proposal — the step that is still done by hand.
Yes, at two gates. A price below the drafting officer's authority is approved by a named manager with a reason code, and every proposal — exception or not — is approved by someone other than the person who drafted it; a proposal the manager drafted routes one level up. The agents never price below list without a reason, never approve, never send and never contact a prospect: they draft, and a person sends. Any agent value can be overridden, and the override takes a reason that stays beside the agent's original figure.
The Agentic Solution Engine builds and deploys it from your own requirements and documents — your service catalog and codes, your list prices and volume tiers, your earnings credit tiers and your authority matrix — and it is live once every quality gate has passed. Most of the setup is loading what you already have: the price list, the authority levels and the clocks your desk is measured on.
Every originator behind the sender screened and measured, not just the sender.
1-min demoTreasury Management Agreement SearchWhat the client agreed, what the log shows, and whether they match.
1-min demoTreasury Management OnboardingLive in days, not a month, with every setup signed and checked.
1-min demoWe'll run Treasury Management Proposals on a sample of statements you have already analyzed, and you can compare what it found with what your team found.