Underwriting, credit appraisal and risk analysis
Every file appraised to the standard of the committee. Every judgement written down as it is formed.
LitFin works a credit file page by page, including the pages that arrived as photographs, asks the questions that file actually raises, and produces an appraisal a board credit committee can interrogate line by line. It says plainly which pages it could not read. The same standard holds whatever trade the borrower is in.
Tailored on a 29 year credit career
The problem
Your best credit judgement is written down nowhere.
It lives with a few people. It transfers by sitting beside them, it is unavailable the moment they are not in the room, and it leaves the institution the week they do. What survives them is a file of outcomes with the reasoning missing, and the reasoning is the only part that could have taught anybody anything.
Meanwhile the file that arrives from a trade nobody in the institution has lent to before gets the thinnest read of the week, because there is nobody to ask. Those are one problem wearing two sets of clothes, and neither of them is a processing problem. Neither is a problem that speed solves.
How it works
Four stages, and the file never leaves your standard.
The same four stages run on every file, in the same order, whoever sent it.
It arrives
Over a signed handoff from your own origination path, or an officer signs in, opens a file here and works it in place. Nothing has to be integrated for an officer to open a file and work it here. The signed handoff is the other route, for institutions that want one.
It is worked page by page
The pages are worked one after another, including the ones that arrived as photographs of paper, in whatever state they actually reached you. The evidence on a page is read where it sits rather than summarised away.
The questions the file invites
The questions come from the evidence in front of them. Where the file calls for it, a climate overlay goes on top of the ordinary appraisal.
The appraisal, and the answer back
A memo folded into the five Cs your committee already works in, and an outcome returned to whoever sent the file.
It does not tire, and it applies the same lenses on a Friday afternoon that it applied on a Monday morning, which is the part a human reader cannot promise and should not have to.
The standard
The career LitFin was tailored on.
Anderson Yohana Mlabwa, LitFin's mentor, spent twenty nine years at one of Tanzania's major banks: eight as a credit officer, then Credit Manager, seventeen as Director of Credit and Secretary of the Board Credit Committee, and two and a half as Director of Risk and Compliance.
He now trains the boards of directors, executive teams and senior credit managers of banks, microfinance institutions and leasing companies. The standard LitFin holds a file to was tailored on that career. That is where the standard came from, and whose judgement shaped the software is the only thing that matters here. You can check that standing yourself.
No logos yet.
Three files. Three sectors. The same five questions asked of each.
Illustrative. These three files are constructed to show that the five headings hold while the answers do not. They are not output from a run, and no customer file appears anywhere on this site.
The headings below are identical. Nothing underneath them is.
| Coffee cooperative, hulling plant | Road haulier, fleet refinance | Hotel, rooftop solar under a green line | |
|---|---|---|---|
| What the money is for | Capital equipment bought once, against a crop that pays once a year. | Existing debt replaced, so the question is what changed since the first lender said yes. | A cost line converted into an asset, where the saving is the repayment. |
| Where repayment actually comes from | Member deliveries and the cherry price, and the cooperative controls neither. | Contracted tonnage and haulage rates, both depending on a small number of shippers. | Avoided diesel and grid spend, which only exists if occupancy holds. |
| What the evidence shows, and what it does not | A member ledger photographed page by page. It shows deliveries. It does not show who else those members deliver to. | Fleet schedules and fuel records. They show utilisation. They do not show the maintenance deferred to produce it. | An installer price quotation and an irradiance estimate. They show a design. They do not show a year of metered output. |
| What the file puts at risk | One season, one price and one buyer relationship, all moving together. | Asset values falling faster than the amortisation that assumed them. | A product warranty that is only as good as the installer still being in business in year four. |
| What could not be read | Pages of the member ledger photographed at an angle the figures cannot be recovered from are named, counted, and excluded from every number. | Two service invoices with no readable date. The maintenance history is reported as partial rather than clean. | The signed page of the offtake annex is missing. The appraisal proceeds and says so. |
The third file is the whole argument in miniature. Green is not an industry we serve, it is a lens that goes over any industry's file. Underneath the overlay it is the same appraisal, asking whether the borrower can and will repay.
Why the institution keeps it
Judgement that used to leave with the person now stays with the file.
LitFin does not sell a course. An officer opens a real file and works it with Mwikila, the AI Senior Banking & Finance Manager, beside them: challenged where the file is weak, asked for evidence where a number is asserted, shown the question a committee will ask before the committee asks it. All of that arrives while the officer is still holding the file, and nobody leaves the desk.
The same pass that produces a defensible appraisal produces a defensible record of how the judgement was formed.
The record belongs to the file
It is a record of a file, never a file on a person. Where the application was prepared elsewhere, the preparation history travels with it. On this side, the appraisal shows its working, so whoever opens the file next can see how the call was formed and not only what it was.
What follows from that
Because the record belongs to the file, there is no index, no overall figure about a person, no rating and no ranking of any officer anywhere in it. A junior can learn from a file with no senior in the room, and a supervisor can see how a call was formed.
The boundary
Every figure it prints, and every observation it makes, can show you where it came from.
Every observation about how a file was worked is dropped unless it traces to something that actually happened. That is the positive half; the list of refusals is published on its own page.
What an engagement is
A pilot is a small number of real files, not a platform migration.
A pilot is a handful of files your committee has already decided, worked beside your own outcome, so nothing live is at risk. It asks for one officer's time, and nothing from your technology team unless you choose the signed handoff.
English and Swahili at full parity, with neither surface ever falling back to the other language. Commercials are set against the shape of that pilot rather than read off a public grid, and the first thing we will tell you is exactly what we need from you before a file arrives.
What a first engagement looks like differs by institution.
Start here
Send us a file you have already decided.
The most useful first conversation is not a demo. Take a file your committee has already decided, in whatever state it actually exists in: photographed pages, a scanned statement, the annex that never got signed, and tell us what you decided and why. We work it, and you compare our reading against the reading you already trust.
A redacted file or a representative sample is fine, and we will tell you exactly what we need before you send anything.
Before you send anything: where the file is held, who can open it, how long it is kept and on what basis it is processed. Data protection and security
Built in Tanzania, expanding across East Africa.