SOSX®, AI-enabled systems thinking, for any sector.
The central book of the SOSX series
SOSX: Seeing the Whole System
The value of systems modelling in the age of AI
Written for anyone who has to reason about a system bigger than one head can hold. It is the general case the products build on.
Seeing the Whole System is the central book of the series. It makes the general case for systems modelling: the theory, the practice, and what agentic AI now changes. It is written to stand on its own. If you read one book of the eight, read this one.
It starts with the cost of not seeing the whole system and ends with an invitation: bring us a real question. In between it covers the discipline (boundaries, feedback, emergence, leverage), then the practice: workshops that produce models rather than murals, parameterising with provenance, red-teaming your own model and earning the right to be believed. The final third turns to the machine: what agentic AI genuinely changes about all of this, and what it must not be allowed to change.
The shape of the book, and of every book in the series, is deliberate: the discipline first, then the practice, and then the machine and what it does for you. Part II is fifteen chapters of method you can run with a whiteboard, a reading list and a spreadsheet: framing, boundaries, stakeholders, provenance and confidence, structured expert elicitation when there is no source, scenario planning, the risk register as a live instrument, decision trees and switching values. Part III describes what agentic AI does with all of it, method by method: what it automates, what it makes auditable, what discipline it enforces, and what stays your judgement. If you never use our software, Parts I and II still do their job.
Each of the seven domains surveyed in Part IV has its own companion book, applying the same discipline to one field in depth.
Inside the book
Part I, The discipline
Chapters 1–7. Where systems thinking comes from, and its working parts: boundaries and hierarchy, connections, stocks and flows, feedback loops, emergence, the iceberg and the archetypes, and the leverage points where a push actually moves things.
Part II, The practice
Chapters 8–22, and the heart of the book: framing the broad question, drawing boundaries honestly, stakeholder power and interest, workshops that produce models rather than murals, parameterising with provenance and confidence, structured expert elicitation when there is no source, normalising unlike things, validating and red-teaming, the failure modes in the room, what-ifs and optimisation, scenario planning, the risk register as a live instrument, decision trees and switching values, reading a model's answers critically, and keeping models alive. All of it doable by hand.
Part III, The machine
Chapters 23–29. Why systems modelling stayed niche, what agentic AI changes, and what it must not, the six-step flow from one broad question to auditable analyses, grounded research and citation discipline, governance as the adoption surface, provable rather than merely plausible, and a method-by-method account of what the machine automates, what it makes auditable, what it keeps honest and what stays your judgement.
Part IV, One spine, many domains
Chapter 30. The same discipline, eight ways, a short survey of each field the companion books then take in depth, closing with an invitation: bring us a real question.
Plus a glossary and further reading in the appendices.
From the series’ shared figure library, the same illustrations you’ll find in the book.
Then go deeper on your own field: browse the eight companion books.