Institutional evaluation
How would we find out whether this works for us?
An evaluation is your institution running real work through RunPayway™, with your own people, and then reading what you are left holding. This page states what you would be testing, what you would supply, what RunPayway™ would do, what you would receive, and how you would judge it — so you can decide whether an evaluation is worth starting before you speak to anybody here.
Financial Fact Verification Infrastructure. You would not be evaluating a lending decision, and you would not be marking a machine's homework. You would be finding out whether RunPayway™ turns the financial evidence you already handle into a verified account of what a household is committed to, and into a record you can still defend months later — while asking less of your people than the way you do it now.
What you would actually be testing.
One claim, put in a form your institution can prove or disprove with its own cases. Everything else on this page exists to let you test it.
That RunPayway™ can take the financial evidence your institution already supports, establish what a household is committed to well enough for your people to stand behind it, issue a record that explains itself without us in the room — and require your staff only where the evidence genuinely does not settle the question.
Not a test of AI accuracy
Nothing RunPayway™ reads is treated as true because a machine read it. A machine reading carries weight only once a person in your institution takes responsibility for it, so a score for how often the machine was right would be measuring a thing the product does not rely on.
Not a test of lending decisions
RunPayway™ issues no approval, no decline, no eligibility and no affordability opinion, so there is no decision of ours for you to agree or disagree with. Your credit judgment stays exactly where it is, and an evaluation does not compare itself against it.
Not a test of an integration
Your staff work in a web application. There is no connector to your core or origination system to stand up, so nothing here tests whether one would have worked — and an evaluation is not blocked behind your engineering roadmap.
What an evaluation looks like, end to end.
Six stages. There is no evaluation mode and no sandbox: from the second stage onward this is the ordinary product doing ordinary work, which is the only version of an evaluation whose result means anything afterwards.
Scope
You decide what would count as an answer
Before anything is set up, your institution names the fact class it wants tested, the kind of evidence it would bring, and what it would need to see to call the evaluation informative. We write that down with you rather than proposing it for you.
What that involves
This is the stage where the absences on the institutional review page matter most. If your process requires a certification, a connector, a single sign-on route or an assurance RunPayway™ does not have today, that is knowable in a conversation and does not need an evaluation to discover. We would rather lose the stage here than in week six.
Your workspace
Your organization is created and your administrator staffs it
An administrator in your institution invites the colleagues who will take part and names what each of them does. Nobody joins by signing up, and nobody at RunPayway™ works inside your organization.
What that involves
Three responsibilities exist: administrator, assessor, and evidence reviewer. Recording what a document says and confirming it against the stored file are deliberately different authorities, so an evaluation with two participants already demonstrates the separation your auditors will ask about. Access can be changed, stopped, ended and restored by your own administrator throughout.
Real work
Your people run real cases through the ordinary workflow
Your staff register real financial evidence on real cases and work them the way they would work them in production. RunPayway™ prepares; your people record, confirm, and resolve what the evidence leaves open.
What that involves
Nothing about this stage is special. The same authorization runs on every request, the same isolation applies to your organization's records, the same refusal to issue on unresolved evidence applies, and every consequential act is written into the same governed history in the same transaction as the change it describes. An evaluation that behaved differently from production would be evidence about the evaluation and not about the product.
Records
Completed records leave RunPayway™ Records you keep
Each completed evaluation becomes a record carrying the result, the method version applied, the evidence it stood on, who confirmed what, and what was left unresolved. You can download it as a file and hand a portable copy to somebody with no account here.
What that involves
The record is the object of the evaluation, not a report about it. It is assembled by the same reader that renders the record on screen, so the file you keep and the page your staff read cannot describe the same record differently — and a completed evaluation can be reproduced from the inputs frozen at the moment it was issued.
Value review
You read what RunPayway™ did and where your people were needed
Inside the product, your organization can read the division of labour over the period: the work RunPayway™ performed alone, the acts that required one of your people, the places RunPayway™ stopped rather than guessing, and the governed output you now hold.
What that involves
Every figure there is a count of governed acts that were already recorded months before anybody asked the question — not telemetry, and not a metric collected for a sales conversation. It reports no rate, no percentage, no trend, no benchmark, no time and no cost, because the history does not record those and deriving them would be an invention with arithmetic in front of it. It never names a person and never reads a borrower.
Your decision
Your institution decides, on its own criteria
You hold the records, the history, and the account of the work. What that is worth, whether it clears your bar, and whether you adopt RunPayway™ are entirely your institution's to determine.
What that involves
RunPayway™ publishes no pass mark for an evaluation and does not score one. We can tell you what the product did; we cannot tell you whether it was enough for you, and a vendor that offered to would be answering a question it has no standing to answer.
What your institution would need to bring.
Four things, and none of them is engineering work. This list is short because the product is used, not installed.
People, with responsibilities you assign
At least two colleagues, because recording a figure and independently confirming it are separate authorities and one person cannot hold both ends of that. Your administrator assigns the responsibilities, changes them, and ends them, inside the product.
Real financial evidence, of a kind the product supports
The documents your institution already collects for the fact class in scope. Real cases rather than constructed ones: a specimen file agrees with itself, and the properties worth evaluating here are the ones that only appear when two documents disagree.
Judgment, where RunPayway™ explicitly asks for it
Confirming a figure against the stored document, choosing between readings that contradict each other, and supplying what a document does not say. RunPayway™ asks only in those places, and where the evidence settles the question nobody is asked anything at all.
Your own policy, applied to the record rather than inside it
What a result means, what your thresholds are, and what happens next stay with your institution. Nothing you tell us changes what RunPayway™ concludes, which is the property that makes the conclusion worth having.
There is no integration to build, no data to migrate, no environment to provision, and no change to your core or origination system. There is also no identity integration: staff sign in with an account your administrator provisions. If your institution requires single sign-on before staff may use an external system, that is a genuine blocker today and is better raised in the first conversation than discovered in the third week.
What RunPayway™ does while you do that.
The same five steps on every case, in the same order, whether it is the first case of an evaluation or the ten-thousandth of a production year.
Prepares
Machines prepare, and claim nothing
RunPayway™ reads the stored documents, computes payments, amortizes terms, and works out remaining monthly margin and stress positions. All of it is arithmetic performed identically every time, and none of it counts as verified because a machine produced it.
Asks
People verify, only where it is needed
RunPayway™ puts a question to one of your people where the evidence does not settle it — an unverified figure, a reading a reviewer sent back, two documents that disagree — and nowhere else. No queue fills because a day passed or a page was reopened.
Governs
RunPayway™ governs what may carry weight
Which figures are current, what superseded what, which are confirmed and by whom, what is still unresolved, and how the methodology treats each one. Facts about what is committed are kept apart from the treatment a versioned methodology applies to them, so a financial fact never changes because a policy did.
Issues
RunPayway™ issues a record that explains itself
A completed result, its hash, the method version applied, the evidence it stood on and what was left open — or a refusal that names what is missing, because RunPayway™ will not conclude on evidence nobody has settled.
Stops
And then stops
There is no field in a RunPayway™ record an approval, a decline, an eligibility, a recommendation or an affordability opinion could be written into. The next step is your institution's, and the responsibility for it stays there.
No change means no work. Where nothing has moved, nobody is asked for anything — which is why the honest measure of an evaluation is not how much RunPayway™ did, but how little of it your people had to do twice.
What you would be left holding.
Not a report about an evaluation. The actual governed output of real work, which stays yours and does not depend on our interface to mean anything.
For each completed record
- The result, its hash, the method version applied, and the instant it was issued
- The documents whose confirmed values were accepted into it, and the figures each established
- What RunPayway™ read for itself, what a person asserted, and what a reviewer confirmed — kept as three different facts
- Which sources superseded which, so a withdrawn document is distinguishable from a current one years later
- What was left unresolved, reported as what is missing rather than absorbed into a total
- The audit references that establish who acted, on what, and in what order
- A downloadable file containing all of it, and a portable copy a third party with no account can check against what RunPayway™ recorded
Across the whole evaluation
Your organization can read record activity across every case at once, administration history separately, and an operating record that adds up the division of labour: what RunPayway™ did on its own, what your people were needed for, where RunPayway™ stopped rather than guessing, and the governed output you now hold. Your administrators can also ask the product whether your institution's recorded history still holds.
Those are counts of governed acts that were already recorded, not metrics collected for a sales conversation. There is no rate, no percentage, no trend and no benchmark in them, and no hours, cost or headcount — the history records that an act happened and when, never how long it took or what it was worth, so those figures are not derivable and are not derived. It never names a person, and it never reads a borrower.
How you would know whether it demonstrated anything.
Six checks you run against your own records, at the end, without asking us. Each is either true of what you are holding or it is not.
Every material number can be placed
Take any figure in a completed record and find the stored document it was read out of and the passage within it. A number nobody can place should not be in there — take a sample and establish whether one is.
Disagreements were surfaced, not smoothed over
Where two of your documents contradicted each other, the record should show the contradiction being put to a named person, together with the readings they chose between — and never a blended figure, and never the most recent reading quietly winning.
Human intervention is visible and attributable
Every place one of your people supplied judgment should be recorded beside the result, with who did it and when. Confirming should be a different person's act from recording, and the record should show that it was.
What is committed is held apart from what was made of it
The account of what a household is committed to should stand on its own, separate from the treatment a versioned methodology applied and from any conclusion your institution drew. Establish that no policy view has leaked into the financial facts.
It stopped where it should have stopped
Look for the cases RunPayway™ refused to classify and read what it said was missing. A product that produced a figure for every case you gave it would be telling you something worse than a refusal.
The record still holds up without us
Reproduce a completed result from the inputs frozen when it was issued. Hand a portable copy to a colleague with no account and have them check it. Read a downloaded record months later and establish that it still explains itself.
RunPayway™ publishes no pass mark, no target, and no percentage any of these must clear. We have no customer benchmark to compare you against and would not offer one if we had. What counts as a demonstration of value is your institution's judgment, made on your own records, against your own bar — and that is the same boundary that governs everything else this product does.
What running an evaluation would not mean.
RunPayway™ measures.
The institution decides.
An evaluation changes nothing about where RunPayway™ stops. Real cases, real evidence and real staff do not move the boundary, and finishing one does not commit either side to anything.
- It does not mean RunPayway™ approved, declined, or priced anybody. No record issued during an evaluation contains a decision about a person.
- It does not mean RunPayway™ recommended a credit outcome, ranked applicants, or offered an affordability opinion.
- It does not mean RunPayway™ underwrote anything, or that your underwriters were replaced or reviewed.
- It does not mean RunPayway™ certified your institution as compliant with any regulation, standard, or framework. RunPayway™ holds no certification of its own and issues none to anybody else.
- It does not mean the results have been validated by any third party, auditor, or regulator.
- It does not mean your institution has adopted RunPayway™, and it does not oblige you to.
- It does not create an integration, an assurance, or a capability that did not exist before it started.
Scope an evaluation in the first conversation.
Bring the fact class you want tested, the evidence you would use, the people who would take part, and what you would need to see. We will tell you plainly whether RunPayway™ can support it today, and where it cannot.
How long an evaluation runs, how many cases it covers, what it costs, and what support it carries are established in writing with your institution and are not published here. RunPayway™ is an early-stage platform: we would rather agree those with you than post a programme we have not run with you.