Formula Inventory Beats Agent Prose
A synthetic Formula Drift control: when an AI finance agent claims an input-only edit, compare the formula inventory before trusting the explanation.
The Variance Check Has To Fail On Purpose
A synthetic budget-versus-actuals test where the plausible finance-agent answer fails four negative controls.
Finance Agent Integrity Pack
A review kit for AI agents that touch financial models: assumptions, scenario state, validation evidence, and audit trail.
Budget vs. Actuals Drift: The Agent Blamed The Wrong Line
A synthetic variance-review test where the plausible agent note blames marketing, while the validated packet points to hosting COGS.
When A Budget Variance Note Sounds Right And Is Still Wrong
A synthetic budget variance note can sound plausible and still be wrong when an agent ignores period boundaries.
The audit trail is the product
Two small protocols for finance agents: leave an audit trail, isolate scenarios, and stop confusing confident prose with model control.
A CFO agent is not autonomous until it can audit its own model
A tool-agnostic protocol for testing whether finance agents can change models without destroying the audit trail.
Can an agent apply a 15% SaaS revenue miss without corrupting the model?
A synthetic Layerz experiment: a 15% revenue miss changes a SaaS cash plan from safe to guardrail breach in month 4.
LeChiffre AI: a public financial-modeling lab
LeChiffre becomes an English public modeling lab: synthetic models, explicit assumptions, validation checks, and limitations.