Skip to content

All courses

Industry track, The AI Decision Course

AI in banking and financial services

For leaders in banks, lenders and financial firms: what AI may tell a customer about their money, lending rules it must sit inside, financial-crime alerts it may close, customers who cannot pay, and when an agent moves the money.

9 lessons, about 3 hours with the exercises, 9 cards your team fills in.

Part of The AI Decision Course: membership is annual and includes all 21 courses, with a full refund within 14 days.

About this course

This course covers bringing AI into banking and finance: putting the model inside your lending rules, gating generative AI like any other bank model, and which financial-crime alerts AI may close on its own. It also covers what the decline letter has to say, what AI may tell a customer about their money, the customer who can't pay, core-banking migrations, the fintech's AI-written code, and when an agent moves the money.

You leave able to wrap a model trained on your own lending history in written policy rules that override any score, to run one gate for every model including the AI inside software you never chose, and to let AI close only the provable crime alerts. You can also route anything that recommends to a licensed advisor and fund only narrow, budgeted agent mandates.

Lessons in this course

  1. Put the model inside your lending rules2-minute read, Free to readEnds in Lending rules card, 20-minute exercise
  2. Gate generative AI like any other bank model4-minute readEnds in AI launch gate, 20-minute exercise
  3. Which financial-crime alerts AI may close on its own4-minute readEnds in Alert disposition card, 20-minute exercise
  4. What the decline letter has to say3-minute readEnds in Adverse-action card, 20-minute exercise
  5. What AI may tell a customer about their money3-minute readEnds in AI escalation card, 20-minute exercise
  6. How far to let AI go with the customer who can't pay3-minute readEnds in AI collections rung card, 20-minute exercise
  7. Which core-banking migrations to hand to AI3-minute readEnds in Core-migration decision card, 20-minute exercise
  8. Will the fintech's AI-written code pass your audit?3-minute readEnds in AI-written code audit record, 20-minute exercise
  9. When an agent moves the money3-minute readEnds in Agent-readiness card, 20-minute exercise

What you leave with

One card per lesson, blank and with a worked example, so the decision is written down before the meeting where it is made.

  • Lending rules card
  • AI launch gate
  • Alert disposition card
  • Adverse-action card
  • AI escalation card
  • AI collections rung card
  • Core-migration decision card
  • AI-written code audit record
  • Agent-readiness card

Back to the course list

Questions

How do we use AI in lending decisions?

Train a predictive model on your own lending history, then wrap it in written policy rules that override any score it gives. Write your experts' instincts down as rules, and know your fully loaded cost per loan before you set the automation budget. The lesson 'Put the model inside your lending rules' covers the shape.

Does generative AI need model risk governance?

Run one gate for every model: the chatbot and the internal summarizer clear the same bar as vendor-embedded software. Count the AI you didn't buy, since ordinary software arrives with AI switched on by default, and keep the sign-off independent. The lesson 'Gate generative AI like any other bank model' covers the gate.

Can AI close financial crime alerts without a person?

Let AI close only provable alerts, the screening hit graded as definitely not the listed person. A person owns the risk on the rest, and you sample what the machine closes to learn when it is biased, hallucinating or gone rogue. The lesson 'Which financial-crime alerts AI may close on its own' covers the tiers.

The other courses in the membership