Solution · Lending · credit for businesses

Every credit application assessed automatically, and only a person declines

We build this system to order for lenders and for companies that sell to businesses on credit. Ordinary software pulls the company’s records, payments and credit report and checks your rules, and AI reads the statements, weighs the payment history and gives the risk in words, with the reasons, and a limit the customer can carry. Low-risk limits under your ceiling are approved by the rules. The rest goes to a credit analyst, and only a person declines

Who works with it

Credit analyst

Decides on credit applications

Gets the limits over the ceiling, the applications AI cannot read and each one AI would decline, with the data, the checks and AI’s reasons

Head of credit risk

Sets how much risk the company takes

Sets the ceiling the rules may approve, the rules themselves and the risk policy. Sees every decision with its reasons

Business customer

A company that asks for credit

Gets a small limit approved quickly. A decline always comes from a person, with the reasons

What comes in

What the system reads for each application, and where the decision goes

CRMthe applicationAccounting datastatementsPaymentson time or lateCredit bureauexternal reportLending systemcurrent limitsThe reviewreads, changes nothingRisk, limit and routeto the lending systemAIAICRMthe applicationAccounting datastatementsPaymentson time or lateCredit bureauexternal reportLending systemcurrent limitsThe reviewreads, changes nothingRisk, limit and routeto the lending systemAIAI
Three ways out

The path of one application

Credit applicationa company asks for a limit
AIThe reviewrules and AI
  1. Approved by the ruleslow risk, under the ceiling
  2. Credit analysta big limit, or AI unsurea person decides
  3. Held for a personAI would declinea person decides
Example applications

Pick an application and see what happens to it

Made-up companies asking for credit. The ceiling and the rules stand for the ones you would set

Pick an example application

A retailer that grew this year asks to double its limit

The risk, in words AI
  1. Low risk
  2. Medium risk
  3. High risk
Asked for$80,000
AI proposes AI$60,000

Rule The rules may approve up to: $25,000

What the system pulled Rule
  • Payments: two years, always on time
  • Statements: revenue up 40% this year, margins steady
  • PassedCompany active: Registered and filing
  • PassedNo overdue debts: None at the bureau
What AI weighed AI
  • SaferRevenue up 40% this year
  • SaferAlways paid on time
  • RiskierAsks for twice its current limit

AI’s reading: can carry more credit, but not double at once

Where the application goes AITo a credit analyst

The limit is over the ceiling the rules may approve, so a credit analyst decides, with AI’s proposal and reasons in front of them

a person decides

Software, AI and a person

Who does which part of the work

Software and AI read every application. A person makes every decision that is not a small, safe yes

Rule

Ordinary software

Monotonous work with clear rules

  • Pulls the application, the company’s records, the payment history and the credit bureau report
  • Checks your rules: the company is active, no overdue debts, the application matches the statements
  • Approves a limit under the ceiling when the risk is low and every rule passed
  • Sends the rest to a credit analyst, with everything the system found
  • Writes the limit, the reasons and the route to the lending system
AI

AI: understand, weigh and choose

The small decisions on every application

  • Reads financial statements and finds figures that do not agree with the application
  • Weighs how a company pays: on time, later each month, or with the season
  • Reads the risk as low, medium or high, with the reasons in words
  • Proposes a limit the customer can carry, which may be less than asked
  • Says when there is too little history to read the risk
a person decides

A person

Every decline and every big limit

  • Only a person declines an application, and the customer gets the reasons
  • The credit analyst decides every limit over the ceiling
  • The credit analyst takes the applications AI cannot read and the figures that do not agree
  • A customer can ask for a decision to be looked at again by a person
  • The head of credit risk sets the ceiling, the rules and the risk policy
How a project starts

Four stages. After each one, you decide whether to go on

First on past applications, then on new ones

  1. Usually firstDiscovery and specification
    In
    How credit decisions are made today, your rules and the data you keep about each customer
    You get
    A written specification: which applications come first, what data the system needs, what it will cost and in what order. The document is yours whether or not we build it
  2. ThenTest on past applications
    In
    Past applications, your decisions and how each customer then paid
    You get
    What the system would have read and decided for each, next to what your team decided and how the customer paid
  3. ThenPilot on one product
    In
    One credit product or one group of customers you choose
    You get
    A risk, a limit and a route for each application. A credit analyst confirms each decision while the rules are being set
  4. When you decideRollout
    In
    The products and customers you choose
    You get
    Small low-risk limits are approved by the rules, and the rest go to an analyst. Your team changes the ceiling and the rules
Six questions we ask first
  1. How many credit applications do you review a month? We plan the system and the pilot around it
  2. Do you know how past customers paid after your decisions? The test on past applications is built on it
  3. What data do you use for a credit decision today? The system starts from the same data
  4. How long does a credit decision take now? The pilot is measured against it
  5. Do you get reports from a credit bureau? If you do, the system reads them; if not, it works from your own data
  6. Which lending or accounting system do you use? The pilot connects to it, and every decision is logged there

Tell us how credit decisions are made today

A 30-minute call. We will say what your application history allows and where a project would start