Blog
Use cases, methods and field notes for finance departments, investment funds and accounting firms.
24 articles
Featured
An AI usage policy sets what is allowed, with which data and who validates the results. Here are the sections it cannot do without.
A successful first AI use case starts with simple questions. Here are the five an executive should ask before launching the project.
Launching an AI plan without a diagnostic often means choosing tools before knowing the problems. Here is the scoping method to insist on.
.webp?alt=media&token=a49e3e09-4703-441a-9454-79f4bcfaf298)
This article explores leveraging AI to generate Python scripts for Excel, specifically for financial analysis tasks like cohort retention. The key advantage is that the code executes locally within the user's secure environment, preventing sensitive data from leaving the system. It covers prompt engineering for secure, commented, and efficient code generation, including data cleaning and error handling.
.webp?alt=media&token=583aca44-f124-47fb-9b7f-84a557e43c4c)
This article delves into the critical aspects of data governance and confidentiality when integrating AI with sensitive financial information. It emphasizes that secure data handling happens *before* the prompt (anonymization, pseudonymization) and through rigorous *validation* of AI outputs, rather than solely relying on AI to 'self-police' sensitive data.
.webp?alt=media&token=995ce30a-fa85-4285-bcf9-71e6ad24a2c5)
This article demonstrates how AI can be a powerful tool for regulatory and tax monitoring. It focuses on using AI's strengths in research and synthesis while rigorously framing its weaknesses. Key strategies involve explicit source citation, structured alert generation, and human validation to ensure legal and fiscal compliance, especially concerning complex regulations like Pillar Two and IFRS.
.webp?alt=media&token=368101f0-689b-4799-9550-3194fcc56f1b)
This article explores how to implement robust audit trails and cross-validation protocols for AI systems in finance, moving beyond mere trust to verifiable, traceable, and compliant AI operations. It focuses on multi-agent architectures and specific prompts to ensure every AI decision is transparent and auditable.
.webp?alt=media&token=5ff68123-18ac-4c3a-909d-d65547c4de24)
As an expert, I analyze the evolution of banking fraud detection, from rigid rule-based systems to AI-augmented behavioral approaches. This article details the concept of the perplexity score for alert prioritization and examines the challenges related to explainability and predictive defense.
.webp?alt=media&token=5b5e6e12-8866-4523-b382-8c70e1bb1574)
A deep dive into the role of Artificial Intelligence in investment, focusing on managing market volatility and optimizing trading timing. As Louis, I break down how NLP and Machine Learning filter noise, detect stress signals, and assist in rationalizing decisions, while covering the structural limits of quant models (overfitting, data quality).
.webp?alt=media&token=6d4ddc4d-7e4c-45cb-a3c6-e0fc9983cbb8)
A complete guide to mastering Perplexity Finance. From configuring requests (prompts) to verifying sources (SEC, reports), this article explains how to transform the tool into a true financial analysis copilot, complementary to brokerage platforms and Google Finance. As Louis, I'll share my experience.
Showing 9 of 23 articles
Return on AI
The newsletter for CFOs, accountants and investment funds: a concrete case, the mechanics behind the buzzword, and AI news filtered for you.
See the newsletter page