Called in when someone suspects a specific fraud already happened and needs it found, quantified, and made to survive a deposition.
Dashiell is an AI Senior Forensic Accountant persona modelled on fraud-examination methods that establishes predication before touching a record, preserves evidence with a chain-of-custody log from day one, runs Benford's Law and statistical anomaly screens before tracing flags to source documents, assesses the fraud triangle (pressure, opportunity, rationalization) for every named suspect, quantifies loss with a stated methodology and confidence caveats, and documents to expert-witness standard throughout, while drawing a hard line at legal conclusions and routing prosecution decisions to counsel.
Who it's for
Controllers, boards, and outside counsel who need a rigorous, litigation-ready investigation into a specific suspected fraud rather than routine control testing.
Key capabilities
- States the predication and evidence-preservation steps before any analysis begins
- Runs Benford's Law, duplicate-payment, and round-number tests, then traces every flag to source documents
- Builds a fraud triangle assessment (pressure, opportunity, rationalization) per named individual, mapped to a specific control fix
- Quantifies loss with a stated calculation method, data period, and confidence caveats
- Writes expert-witness-ready reports with fact separated clearly from opinion, and never renders a legal conclusion
How to use it
Paste Dashiell's SKILL.md into your Claude Project Instructions (or any AI system prompt), then describe your suspected fraud, anomaly data, or litigation-support documentation problem. Works with Claude, ChatGPT, and any AI chat. Under 2 minutes to install.