Reviewing With Machines
AI in credit review, taught for practitioners: what the machine is, where it fits, how it fails, and what has to be true for AI-assisted work to survive an examiner. The modules build on one another — start at M1 and take them in order. About four hours, self-paced, free and unwalled, and you leave with artifacts you can take to your own governance committee.
AI can produce work. Only a human can own a conclusion. In credit review, the AI is never the accountable party — not for a risk rating, not for an issue, not for a sign-off.
The beginner-friendly foundation. Start here to understand the mechanics, fit, structured prompting, and human accountability.
- M1
Start with one imperfect AI draft
Learn the basics through a short borrower example: what AI can produce, what it does not know, and how to label facts, implications, gaps, and questions.
- M2
Where AI fits in the workflow
A practical framework for separating tasks AI can accelerate from tasks that require a human accountability moment.
- M3
Give AI a well-defined assignment
How to ask for useful work without inviting unsupported conclusions. Build a source-bound, constrained assignment using the ASSIGN framework.
- M4
Check the work before using it
Apply TRACE to find numeric, source, interpretation, omission, and ownership errors before polished AI output enters the workpaper.
- M5
Own the conclusion (The RACI)
The centerpiece. Who is Responsible, Accountable, Consulted, and Informed for every task in the lifecycle — and the one assignment that is locked by construction.
- M6
Mini case — one borrower, one review task
Integrate task fit, ASSIGN, TRACE, and human ownership in a bounded review of fictional Northstar Industrial Supply.
Deep dives into specific review tasks, with fictional files and validated prompt libraries.
- A1
Track A: Documents and Covenants
Extracting dates, summarizing terms, and preparing compliance checks.
- A2
Track B: Financial Analysis
Initial spreading, ratio recalculation, and flagging add-backs.
- A3
Track C: Risk Identification
Scoping analytics, sampling methodology, and identifying potential issues.
For review managers and program owners. Verification tiers, bias, regulation, and standing up an institutional pilot.
- V1
Automation bias and verification tiers
Why a polished draft suppresses challenge, why the second reviewer defers to the first machine, and why “I checked it” degrades to “it looked right.”
- V2
Governance and regulatory context
AI-assisted review inside existing supervisory expectations — model risk management, loan review guidance, third-party risk, fair lending, and the emerging AI frameworks.
- V3
Standing up AI in your own shop
Pilot design, scope limits, a challenge log, the first 90 days, metrics that actually detect degradation, and when to switch it off.
- V4
Triage-based coverage — designing escalation, not just prompts
Full-portfolio coverage with risk-allocated human depth: tier design, deterministic escalation triggers, the random-audit control, and the metrics that make continuous review governable.