FOR LENDERS

Credit decisioning on the whole loan file.

Floowed reads every document an application arrives with, whatever state it is in, screens it for tampering, computes what the account actually did, finds the debt nobody declared, and then runs your credit policy on all of it. Lending is where we started, and it is still the deepest thing we do.

One loan file · Application #4402Decided
  1. 01Arrived
    • Application, from origination14 fields
    • Bureau report, pulledReceived
    • DocumentsIDs, 3 statements, passbook, payslips
  2. 02Read
    • Bank statements, 3 banks12 months
    • Passbook, photographedRead
    • Payslips and IDsRead
  3. 03Screened
    • Typography, arithmetic, pixelsNo signal
    • Against the applicationConsistent
  4. 04Analysed
    • Average daily balance412,300
    • Bounced cheques and NSF2 events
    • Negative days4
  5. 05Discovered
    • Repayment streams3
    • Declared on the application2
    • Undeclared1
  6. 06Bureau
    • Report processed2 open accounts
    • Enquiries, six months1
  7. 07Checked
    • Entity, against the registryActive
    • Sanctions and PEP, 5 subjectsNo match
  8. 08Credit policy
    • Hard gates7 of 7
    • Scorecard78 · Grade B
  9. 09Written back
    • Decision, to the LMSWritten
    • Evidence kept with the fileYes
Approve to the requested limit. Grade B, 7 of 7 gates.Recommended, a credit officer decides

SEE IT RUN

A loan application, start to decision, in two and a half minutes.

One application through the whole path on the real product: the documents read, the statements checked, the undeclared debt found, the credit policy run, and the decision explained line by line.

RUNNING ON FLOOWED

  • Alon Capital
  • Discovery Capital
  • Yulon Finance
  • Gedit
  • ASPAC Bank
  • Kredit Hero

THREE WAYS IN

A credit decision draws on three sources. Floowed takes all three.

Most applications use more than one, and which ones you use is your choice, not ours.

Your systems

Your LMS, your core banking platform, the origination front end, and whatever else already holds the application.

40+ integrations across the lending stack, plus an API that takes a case in and writes the decision back

The bureaus and registries

Credit bureau files, company registries, KYC and identity providers, sanctions and PEP lists, and banking data providers.

Pulled at the point in the policy that needs them

The documents

Statements, passbooks, financials, filings and the borrower's own papers, including the scanned, photographed and handwritten ones.

Uploaded, emailed, or watched in a folder

If the data is already in your systems, nobody ever has to touch a document. If it arrives as a photograph of a passbook, that is not a problem either. The same credit policy runs on whatever the file is made of.

WHAT HAPPENS TO THE FILE

Everything below happens before a credit officer opens the application.

Six passes over the same file, each one feeding the policy that decides it. What comes back is a graded recommendation that shows its work, in minutes rather than the days a manual review takes.

DOCUMENT INTELLIGENCE

It reads the file the borrower actually sent.

Loan files do not arrive clean. A passbook photographed at an angle, a scan of a scan, a handwritten cash book, twelve months of statements from three different banks in three different layouts. Floowed reads all of it into structured, validated data. Every transaction is replayed against the running balance, and where it does not reconcile we surface it rather than healing the number to make it work. Every figure sits beside the page it came from, so your credit and risk teams verify what was read instead of trusting it.

  • Bank statements, passbooks, financial statements, tax returns, payslips, permits and filings
  • Scanned, photographed, faded, skewed, mixed-language, password-locked and multi-account
  • Statements from several banks normalised into one view of the borrower
  • Every transaction replayed against the running balance, discrepancies surfaced not corrected
  • Every value traced back to the page and the line it came from
Savings passbook · photographed, 3 pagesRead

Page 2, at an angle

DateLineAmountBalance
03 MarDeposit p.2 · l.11+25,00061,450
07 MarWithdrawal p.2 · l.12-8,00053,450
12 MarDeposit p.2 · l.13+14,20067,650
15 MarWithdrawal p.2 · l.14-3,50063,150
21 MarDeposit p.2 · l.15+9,80072,950

Line 14 does not reconcile. Surfaced, not corrected. Every value traced back to the page and the line it came from.

DOCUMENT FRAUD

The statement that balances because somebody made it balance.

Every document is screened for tampering as it is read, on the file exactly as submitted. A doctored statement usually reconciles perfectly, which is precisely why a person reading quickly passes it, and why the check belongs in the same pass that produced the data. Floowed looks at the typography, the arithmetic, the file's own metadata and the pixels themselves, then reports what it found and the page it found it on. Nothing is rejected on a fraud signal alone: a person sees the finding and decides.

The same checklist runs on every document, and most of them come back like this one. What matters is that it ran, and that the result is on the file.

BANK STATEMENT ANALYSIS

What the account actually did.

Twelve months across every bank the borrower uses, turned into the cash-flow view your credit policy needs. Average daily balance, inflows and how concentrated they are, volatility, negative days, and debt service coverage, all computed from the statements rather than taken from the application form. Credit and debit run-rates sit beside the balance, so a month that looks healthy on its closing figure does not get to pass as one.

  • Average daily balance with credit and debit run-rates, month by month
  • Inflow concentration, volatility, and the days the account spent negative
  • Negative-balance episodes tracked end to end: how deep, how long, and whether it genuinely recovered
  • DSCR and coverage computed on the statements, not on what was declared
  • Extraction accuracy, document quality and balance verification reported per statement
Statements · 12 months, 3 banksComputed
Extraction accuracyPer statement
Document qualityLegible, skewed
Balance verified27 of 27 pages

Average daily balance, month by monthNegative days in pink

  • Average daily balance412,300
  • Credit run-rate, monthly1.28M
  • Debit run-rate, monthly1.19M
  • Inflow concentration38% from one payer
  • VolatilityModerate
  • Negative days4, in one episode
  • Deepest, longest, recovered-18,400 · 3 days · Yes
  • DSCR1.6×

Computed from the statements rather than taken from the application form.

BOUNCED CHEQUES AND NSF

The cheque that came back, and the one that came back twice.

Returned cheques and failed debits are the clearest repayment-conduct signal in a loan file, and they are the easiest to miss reading line by line. Floowed separates a bounced payment from a reversed incoming receipt, groups the related cheque events together so an attempt and its reversal are read as one thing, and orders every event correctly across statements even when they arrive out of sequence or their dates overlap. Each event opens onto the transactions underneath it, with the balance before and after, so a DAIF entry is a finding with evidence rather than a number in a summary.

  • Bounced payments separated from reversed incoming receipts
  • Related cheque events grouped, with the attempt and the reversal shown together
  • The balance before and after each event
  • Events ordered correctly across overlapping and out-of-sequence statements
  • Every event drills down to the underlying transactions
Returned items · 3 statements, 2 events2 events
Cheque 0417 · 45,000Returned · DAIF
  • 12 MarPresented-45,000
  • 13 MarReturned, DAIF, reversal+45,000
  • 19 MarRe-presented-45,000
  • 20 MarCleared

Before 38,200After -6,800Recovered 20 Mar

Auto-debit · 12,500Returned twice · NSF
  • 02 AprAttempted-12,500
  • 03 AprReturned, NSF, reversal+12,500
  • 09 AprAttempted again-12,500
  • 10 AprReturned, NSF, reversal+12,500

Before 9,100After 9,100Recovered No

Incoming receipt · 30,000Reversed by the payer · not a bounce
  • 17 AprReceipt reversed-30,000

Every event drills down to the underlying transactions, with the balance before and after.

LOAN DISCOVERY

The debt on the form is rarely the debt on the account.

Borrowers under-declare what they owe, usually without meaning to. Floowed reads the repayment streams actually running through the account and lays them out lender by lender: the typical repayment, how regularly it hits, the monthly equivalent, whether the stream is still live, and the month-by-month record of which payments landed and which failed. Repayments are linked back to the loan proceeds they belong to, so an existing facility is visible from both ends. What was declared is then held against what is servicing, and the gap becomes a gate or a score you control rather than a note somebody might catch.

  • Every repayment stream by lender, with cadence, typical amount and monthly equivalent
  • Repayments linked back to the loan proceeds they belong to
  • Live facilities separated from settled ones
  • Failed repayment attempts shown on the timeline beside the successful ones
  • Declared debt held against the obligations actually servicing out of the account
Repayment streams · 3 lenders1 undeclared
Bank term loanLive

Typical 18,750Cadence MonthlyMonthly equivalent 18,750Proceeds 900,000, June

Fintech working capitalLive

Typical 4,300Cadence WeeklyMonthly equivalent 18,600Proceeds 150,000, OctoberNot on the application

Cooperative loanSettled

Typical 2,900Cadence MonthlyEnded January

Declared 2 · Servicing 31 undeclared

Declared debt held against the obligations actually servicing out of the account.

BORROWER AND ENTITY CHECKS

The file checks out. Do the people?

A business loan is a loan to a company and, in practice, to the people who own and run it. Floowed verifies the entity against the registry, that it exists, that it is active, and that its registered particulars are what the application claims, then reads the ownership off the filings and matches it to the applicants, signatories and guarantors. Those people are then screened against sanctions and watchlists and for politically exposed person status, at the point in your policy that calls for it.

Every check is recorded on the file with what it returned and when it ran, so your credit and risk teams assess a finding rather than inherit a flag. A screening hit is a fact on the application, not a verdict on it, and you set what it is allowed to do.

  • Entity verified against the registry: existence, status, and registered particulars
  • Ownership and beneficial owners read off the filings and matched to the applicants
  • Sanctions and watchlist screening on the company, its directors and its owners
  • Politically exposed person status checked on every subject, and recorded on the file
  • PEP by association covered too: the relatives and close associates of an exposed person
  • Adverse media and open-web research on the company and the people behind it
  • You decide what a hit does: reject, score, or route for review

THE DECISION ENGINE

Your credit policy, running itself.

Everything above exists to feed this. Your credit and risk teams encode the policy once: the hard gates, the thresholds, the scorecard weights, the grade bands, and what each combination of grade and gate is allowed to do. Then it runs, the same way, on every application, on data machine-read from the file rather than re-keyed from it.

An application that clears cleanly can be decided automatically, once you trust it to be. Everything else routes to a credit officer with the analysis already done, every hard rule shown with the value it observed and the condition it was tested on, and every scorecard metric showing what it contributed.

SME term loan · Application #4402Decided · v7
Approve to the requested limit. Grade B, score 78.Recommended

Hard ruleObservedCondition

  • Months of bank data12≥ 6Pass
  • NSF events, 12 months2≤ 3Pass
  • Negative days4≤ 10Pass
  • DSCR1.6×≥ 1.25×Pass
  • Leverage2.1×≤ 3.0×Pass
  • Negative equityNoNonePass
  • Ticket size1.2M≤ 2.0MPass

Scorecard metricPoints

  • Cash flow · 40%31 of 40
  • Repayment conduct · 25%20 of 25
  • Financial strength · 20%15 of 20
  • Relationship · 15%12 of 15

Decided on the policy version live at the timeEvery gate and its evidence on the file

WHAT A POLICY IS MADE OF

Written by your credit team, and changed by them too.

A separate policy version per product, sector or ticket band, each one governed the same way.

  • Hard gates

    One condition each, on any value in the file. Minimum months of bank data, NSF and negative-day tolerances, DSCR and coverage floors, leverage and negative-equity checks, sector and ticket-size rules.

  • Scorecard

    Weighted attributes producing a score and a grade. You set the weights, the bands, and what each band is allowed to do. Bring your own model if you have one.

  • Outcomes

    Not one verdict. Every combination of grade and gate gets its own, and each can carry conditions: approve to the requested limit, approve at half with a review in 90 days, or route to a credit officer.

  • Back-tested before it runs

    Replay a policy change against your historical book and its real outcomes before it goes live, so the approval-rate and NPL impact are known rather than guessed.

  • Every version kept

    Only a person with permission can publish. Every version is kept with who changed what and when, so an application decided in March is still explainable in November against March's rules. Applications in flight finish on the version they started on.

  • The decision record

    Every gate with the value observed and the condition it was tested on, the evidence behind each, the figures worked out along the way, and the policy version that was live when the application ran.

See how a decision is built

HOW IT FITS

It sits between what you already run, and replaces none of it.

  • Your LMS stays your LMS

    Floowed decides. It does not originate, service, or move money. The decision goes back to the system that already runs your book, and nothing there has to change.

  • 40+ integrations across the lending stack

    Core banking, loan management, credit bureaus, KYC and identity, and banking data providers, plus an API in both directions for whatever is not on the list.

  • Live in days, not quarters

    Serious decisioning without the implementation tax other platforms carry. No consulting programme, no year-long build, no engineering tickets on your side.

OPERATING PROOF

The first 90 days at one lender, on their real files.

Scoped deliberately: these are one customer's results on their own book, not a claim about yours.

  • 54%

    Review time on clean applications

    Alon Capital
  • 3×

    Document fraud caught

    Alon Capital
  • 180+

    Officer hours returned weekly

    Alon Capital
  • 0

    Engineering tickets raised

    Alon Capital

Our team used to spend hours per application: reading bank passbooks, checking numbers by hand, reconciling documents. Now Floowed reads the documents, runs our credit policy, and surfaces a decision in minutes.

Rene de JesusFounder, Alon Capital

QUESTIONS

What lenders ask first.

  • Does Floowed replace our loan management system?

    No. Floowed is the layer that works out what should happen, not the system that records it. Your LMS or core banking platform stays the system of record. Floowed takes the application, gathers and checks the evidence, runs your credit policy, and writes the decision and its reasons back.

  • What if our data is already structured and we have no document problem?

    Then you never touch that part of the platform. Applications can arrive entirely through the API and your integrations, and the same credit policy runs on them. Document intelligence is there so that no application is undecidable, not because every lender needs it.

  • Can Floowed read bank statements from several banks in one application?

    Yes, in any format and any quality, including scanned, photographed and password-locked files. Statements from several institutions are normalised and reconciled into one view, so the cash-flow picture covers every account the borrower banks with rather than the one they chose to send.

  • Does it decide on its own?

    Only as much as you let it. Most lenders start with it recommending on everything, with a credit officer approving. Once the recommendations have come back right often enough, you can let it decide the applications where every gate passes cleanly, or everything under a threshold you set. Anything with a failed gate or a conflict still goes to a person.

  • How do we change the credit policy without breaking the book?

    Back-test it. Replay the change against applications you have already decided and compare, so the approval-rate and NPL impact are known before it runs. Publishing is permissioned, every version is kept, and applications already in flight finish on the version they started on.

  • How long does implementation take?

    Days, not quarters. There is no consulting programme and no engineering work on your side. Your credit and risk teams describe the policy, read the draft back, test it against decisions you have already made, and publish it.

Run a real loan file through it.

Start free and put a live application through it, or book a demo and we will run one with you and show the evidence behind every check.