Anyone who has run an IT organisation over the past two years knows the situation. The method is in place, Scrum is running, there is a Kanban board on the wall, the OKRs have been written. And yet something no longer fits. Coding agents produce code that goes to production. Code volume grows faster than review can absorb it. The classic junior position, the one where the craft used to be learned, has suddenly become hard to justify. Most management models have no answer to this, because they were written before the agents arrived. That gap turned into a book, published this week: Hepta Management System, the operating system for your IT organisation, from Lean-Agile to Lean-AI.

📌 The framework was never the problem

The transformation work of recent years kept surfacing the same pattern. An organisation picks a framework, rolls it out properly, trains people, pays for certifications. Two years later the lead time from request to deployment is unchanged, and the search for the next framework begins. Not because the wrong framework was chosen, but because what got copied was what can be seen: the meeting format, the planning tool, the org chart.

The book gives this confusion a name, and that name carries the whole text: the distinction between an action and the content of an action. A child playing at driving turns the wheel, sounds the horn and puts a foot down. The action is reproduced perfectly and the content is entirely absent. Toyota spent three decades refining a system of thinking, leadership and problem-solving culture; the West copied the Kanban cards and then wondered why the results failed to appear. The same pattern can be traced through Scrum, through OKR, through SAFe. Once you know that lineage, telling a pragmatic recommendation apart from a best practice elevated to religion becomes straightforward.

Right now the pattern is repeating itself most clearly around AI. An organisation buys tool licences, sends its teams on a prompt course, and has thereby completed the visible action. Whether value actually lands is decided elsewhere: in requirement clarity, architectural ownership and the discipline to check every line an agent hands over.

📌 Seven senses instead of another framework

An operating system does no work of its own. It makes everyone else's work possible by abstracting complexity and offering one consistent interface. That is the role intended for the Hepta Management System. It replaces neither Scrum nor ITIL nor DevOps. It sits as a layer above them and names seven senses through which an organisation perceives its own work:

• Direction: where are we steering, and what are we deliberately dropping? (OKR)
• Leadership: how are decisions made and delegated? (Management 3.0, Lean leadership)
• Flow: where is value waiting rather than moving? (Lean)
• Cadence: which rhythm suits this kind of work? (Scrum, Kanban, SAFe)
• Stability: does operations run reliably without tipping into bureaucracy? (ITIL)
• Collaboration: how does an idea reach production? (DevOps)
• Reach: how do we extend our own delivery capability? (sourcing, contract models)

Each sense comes with exactly one question that matters. The value lies less in the answers than in directing attention to wherever an organisation currently is not looking. Focus only on agile delivery and you will ship quickly until production falls over. Tend only to stability and you will reliably operate software nobody needs any more. No sense outranks another, and none works in isolation. What emerges is therefore a repertoire rather than a checklist: frameworks become interchangeable without losing orientation.

📌 The AI thread running through the book

Eight chapters form a continuous thread and carry the seven senses into the world of agents. They start with the anatomy of a coding agent and run through teams in the AI era, spec-driven development and the Hybrid Implementation Hypothesis to operations, guardrails and the forensics of AI-induced incidents.

The intention throughout was to work with data rather than with attitude, and the data contradicts both of the popular narratives. The miracle-cure story does not survive scrutiny: the first randomised trial with experienced developers shows people who feel twenty per cent faster and are in fact nineteen per cent slower, a self-perception gap that no better model closes. The soon-forgotten-hype story survives no better: in a quarter of the most recent Y Combinator startups, ninety-five per cent of the code comes from language models. At the same time the code grows faster and gets worse on average, with more duplication, less refactoring and more vulnerabilities. The bottleneck moves from writing to reviewing, pull request volume rises sharply, and the talent pyramid is measurably tipping, because the entry-level tasks on which juniors used to learn the craft are precisely the ones disappearing.

That is the real management task of the next few years, and it does not sit in the tooling. It sits in the question of whether an organisation owns the way of working in which these tools become productive. The DORA research puts it plainly: AI is an amplifier. It magnifies existing strengths and existing weaknesses alike.

❓ How the book is structured

The structure follows the life cycle of an IT organisation rather than the chronology of method history. After a prologue that establishes where the methods came from and sets up the action-versus-content distinction, six parts work through practice. START covers setting the organisation up, with OKR, leadership, team design and the responsibility triad of Value, Flow and Architecture. BUILD covers delivery, estimation, scaling, portfolio, architecture, make-or-buy and spec-driven development. RUN moves into operations, from DevOps and ITIL through cloud economics to monitoring, on-call and securing agent-supported systems. IMPROVE covers the feedback loops, SCALE covers reach through sourcing, contract models, time zones and provider management. RECOVER then deals with what to do once something has already gone wrong, using the ACC principle borrowed from New Zealand's no-fault accident compensation scheme as its ordering frame: stabilise first, then analyse, then improve.

Every chapter closes with a section naming the limits of what it has described. That was a deliberate choice. A management book that only explains when something works is worthless as a basis for decisions. The interesting information usually sits in where an approach stops holding. OKR solves no cultural problems and replaces no strategy. Kanban steers the flow, not the direction. ITIL is no substitute for technical competence, and its greatest danger is not too little of it but too much. Kaizen does not help when a system is fundamentally set up wrong and needs radical change rather than small steps.

📌 Who the book is for, and who it is not for

It is written for CIOs, CTOs, IT leaders, architects and team leads who treat frameworks as tools rather than as matters of belief, and who want to know what endures and what is currently shifting. Anyone looking for certification preparation is in the wrong place: the Scrum Guide and the ITIL Foundation curriculum exist in their original form and serve that purpose better. The same applies to anyone expecting a technical guide to fine-tuning language models. The book treats AI from a management perspective, not an engineering one.

A second caveat belongs here. This is not a method you introduce once and then own. The epilogue asks the question that comes first in practice: where do you start when seven dimensions are in front of you and the capacity for seven simultaneous initiatives is not? The answer cannot be delegated, neither to consultants nor to a tool nor to a newly created role. Addressing the weakest sense first rather than the most interesting one builds momentum without overwhelming the organisation.

📌 Editions and availability

The book is published by HeptaMS Press in two editions, German and English, each as a colour-print paperback and as an e-book. The German edition runs to 542 pages, the English one to 522. The English paperback is priced at USD 69.99, the e-book at USD 33.00, with the equivalent list prices set per market.

The ISBN of the English print edition is 978-1-0672937-2-7, and of the German edition 978-1-0672937-0-3, which means the title can be ordered through any bookshop alongside the usual online channels. All sources are collected on heptams.org, where edition and format can be chosen without first working through regional logic.

If your organisation is currently working on one of the seven questions, your perspective on it is welcome. Every organisation has to find the answers with its own people and its own means; the book supplies the questions and a sober view of the tools others have used to answer them.