Neusler Lite
Issue 02

The Model Is the Bank

What happened when financial institutions put language models into production — in underwriting, compliance, fraud, and the call centre.

Read the issue
The Model Is the Bank cover

In this issue

  1. Underwriting After the Language Model

    Lena Vogel

    Credit decisions are being made on documents no human will ever read.

  2. What an LLM Can and Cannot Do in a Bank

    Devin Rao

    A practical boundary, drawn from two years of deployments.

  3. The Compliance Copilot

    Nadia Farouk

    The most boring AI use case in finance is also the most profitable.

  4. Synthetic Data and the Model Risk Committee

    Sarah Lindqvist

    How do you validate a model trained on data that never happened?

  5. Fraud Models That Explain Themselves

    Hannah Cole

    A decline you cannot justify is a customer you are about to lose.

  6. Interview: Building an AI Risk Function from Scratch

    Priya Menon

    She was employee one. Eighteen months later the function has thirty people.

  7. Customer Support Was the Easy Part

    Omar Haddad

    Deflection rates went up. Then the hard conversations were all that was left.

  8. Vector Search Meets the Transaction Ledger

    Ravi Iyer

    Semantic retrieval over financial records is useful, and quietly dangerous.

  9. The Cost Curve Nobody Budgeted For

    Tomas Alvarez

    Inference is cheap per call. Nobody modelled the number of calls.

  10. Model Drift in a Credit Cycle

    Maya Okonkwo

    Every credit model is trained on a period that has already ended.