Applications of Agentic AI in Financial Mathematics

Building reliable AI agents for Financial Mathematics

Why · What · How

The philosophy behind this book

From the Foreword.

Why

When both stones of a grinding mill move, nothing can be ground effectively. For the grinding to happen, one stone must remain firm and steady, while the other rotates against it. That is the philosophy behind this book. Trying to learn Financial Mathematics and Agentic AI at the same time can make the learning process unnecessarily difficult — it is moving both stones at once. We keep Financial Mathematics as the strong, stable foundation, and let Agentic AI be the rotating stone that moves across it.

Who it's for

Written for a reader who knows some Financial Mathematics — interest theory at the level of a first actuarial or finance course — and can write and run short Python programs. No experience with language models or agent frameworks is assumed.

How it's built

Part I builds the minimum conceptual foundation needed before anything else makes sense — what a language model actually does. Part II introduces the vocabulary and mechanics of agentic systems using general-purpose examples, before any Financial Mathematics is involved: memory, tools, reliable configuration, composability through MCP, and multi-agent systems. Part III is the core of the book — across nine chapters, each one carries a Financial Mathematics topic from formula to a working Agno agent: concept, mathematics, worked examples, Python implementation, agentic application, and where the approach can fail. Part IV addresses what changes when an agentic application moves from a notebook experiment to something intended for real use — architecture, evaluation, guardrails, and observability.

Four parts, twenty-one chapters. Pick a part to see its chapters.