Appendix E5 min readEdited 16 November 2026

How the published theses developed

A connected argument over several years

The reading map connects earlier publications with this book’s themes. The technical chapters and protocol passport specify their respective document snapshots.

ThreadPublicationsContribution
Technology and creativityYin and Yang; early shopping and multimodal articleshuman value and experience
Operational reliefJanuary 2025 whitepaper; Agentic eCommercemerchant work and complexity
From clicks to intentCheat Sheet; Thought-Chain; Customers Turn Into Codedelegation and journeys
Discovery and distributionSoftware Ate the World; Discovery Layerchanging selection
DifferentiationMarket Reset; adaptive storefrontsproduct value and commodity trap
Open AI and development140,000 Lines; When Code Writes Code; Commerce Benchmark; Open-Weight InflectionOpen weights, customization and evidence
PrototypesACP with PayPal; generated storefrontsboundaries and inspectable results
Shared architectureStrategy Book; Developer Guideprotocols, trust and operations
Executable systemsB2B interview; Execution Layerbusiness truth and permissions
Procurement experimentProcurement Swarmbounded end-to-end case

Further early writing on AI shopping, multimodality, spatial commerce and virtual try-ons appears as historical reading in the bibliography. Articles on AI-assisted development and the commerce benchmark add an engineering perspective. Historical promises and benchmark results are not carried forward as evidence of current product capability. [S71–S78]

The engineering strand in the publications

My March and April 2025 essays discuss AI-assisted development through project work. The June 2026 commerce benchmark and July’s “The Open-Weight Inflection” add open models and local execution. They inform Part V’s question: which forms of customization become economical and remain maintainable? Project accounts and research on development productivity provide different kinds of evidence. [S76, S77, S78, S91, S118]

A reading map of the public experiments

The experiment boxes describe the task, setup, finding and status. References lead to the repositories and versions used. This map supports selective reading. [S99–S116]

Project groupTopicChapter
ATELIER / YOU and StoryfrontPersonal presentation and product stories6
AXPExperience within the assistant6
WebMCP and SwagA2UITools and interfaces in the customer channel9
Sales Agent HarnessShop functions for several access routes10
Merchant Quote Agent and rfq-benchNegotiation within agreed boundaries17
Shopware × Claude and Starter KitProposals, approval and execution11
x402Paid digital services8
Agent in a BoxDevelopment environments for each project13 and 16
SwagUcp and official pluginMaintained protocol integration19
Nyx and VEYNDOPersonal experiments with knowledge and adaptation11 and 15

“After Turing” explores agent identification through behaviour or metadata. Identifying a request and authorising a purchase are different tasks. [S105]

Thinking ahead. Building ever since.

The path to this book begins with a question we have explored at Shopware for years: how do AI, commerce and experience fit together? Early ideas become articles, a whitepaper and, eventually, experiments with complete purchasing journeys.

June 2023 · Agents buy. People experience. In two articles, I describe AI agents for routine purchases, personal buying assistants and individually composed shopping experiences. I already see these as connected developments: agents take work off our hands, while people choose to spend time with an offer that interests them. [S169, S170]

21 December 2024 · The shop remains the foundation. Agents coordinate tasks through APIs; specialist commerce systems run the business behind them. That division of work becomes a central architectural question for agentic commerce. [S16]

17 January 2025 · Agentic eCommerce. In “Agentic eCommerce: Unlocking Next-Generation Differentiation”, I connect specialist agents, self-optimising stores and new opportunities to stand out in the market. [S15]

21 January 2025 · The first whitepaper. The dated original version develops these ideas in depth: AI autonomy, merchant operations and competition become a shared strategic challenge. [S01]

Figure 39. The early publications connect routine purchases and experience with an architecture and a broader strategic view. Sources S169, S170, S16, S15 and S01.
Figure 39. The early publications connect routine purchases and experience with an architecture and a broader strategic view. Sources S169, S170, S16, S15 and S01.

Putting the ideas to work.

1 October 2025 · Shopware, ACP and PayPal. I present our ACP implementation for Shopware and a working PayPal prototype. We can now test an agent-driven purchasing journey in practice: selection and conversation meet ordering and payment. [S20]

2026 · Work in the Shopware Agentic Commerce Lab. We develop and test shopping agents, open protocols and new merchant tools. Ideas become reference implementations, prototypes and insights we share with others. Chapter 17 explains how we approach that work. [S116]

October 2026 · From idea to book. The early theses, current developments and our Lab work come together in this handbook: 21 chapters on commerce strategy, technology and leadership, available in full in German and English, free to read online and as a PDF.

Figure 40. Practical work connects implementations, open Lab experiments and the handbook. Sources S20 and S116.
Figure 40. Practical work connects implementations, open Lab experiments and the handbook. Sources S20 and S116.

The market makes this division of work tangible, too. In April 2025, Amazon introduces Buy for Me as a limited beta for US customers: after their confirmation, an agent buys on a merchant’s website. With ACP, introduced in September 2025, the agent connects buyer and merchant; the merchant handles the order and fulfilment. To me, this supports the strategic direction: agents become a new route into commerce, and a capable commerce core becomes more valuable. [S171, S25]

How to read this book

The chapters connect my strategic arguments with published developments, personal experiments and practical decisions. “From the Lab” identifies projects from Shopware’s Agentic Commerce Lab; “From my experimental work” identifies my personal projects. The boxes describe the question, what we built, its main lesson and the next step. They provide a way to apply the ideas to your business and explore the code in more depth.

Nordstern, Werkpunkt and Lichtfeld are fictional merchants. They help us explore outdoor advice, spare-parts procurement and room design. The two anonymous real-world cases in Chapter 16 come from my merchant network. Model calculations state their assumptions; quotations and published facts refer to the sources in the appendix. That is also where you will find protocol versions.

Image credits for this edition

The 21 chapter vignettes form one conceptual visual story with the traveller, suitcase and compass agent. They are illustrations rather than captures of real projects. The Chapter 11 image moves to the merchant’s side: the merchant sets objectives and boundaries for the shop operator.

The author photograph and historical publication covers come from my website. Twelve original screenshots show the public projects discussed, Shopware documentation, and the Storyfront, Agentic Commerce Alliance and Readiness Scanner websites. Captions identify their provenance and, where documented, the capture date or project version. The team photograph comes from Shopware’s official team page. Diagrams are original explanatory models. Numerical graphics show the explicitly labelled calculations or dated published findings.