Learning × resources

Authors

Pierluigi Pugliese and Dario Zanotto, Connexxo. German edition by Dr. Jürgen Hoffmann.

Format

Book, PDF, 103 pages. Connexxo GmbH, 2026; published in English, German and Italian.

Material

Bought through the Connexxo shop. Each PDF is personalised to its buyer, which is one more reason it is not reproduced here; characterised only.

An evidence-based account of how teams of humans actually work with AI agents, positioned against both marketing poles — the claim that nothing essential has changed, and the promise that human teams are about to disappear. It runs in six parts: what the evidence supports on team size and on the patterns of human-agent collaboration; how teamwork, developer skills and the path from junior to senior change; the technical disciplines that keep agents on course, from the XP practices through domain-driven design to shared standards and the platform; the hidden costs — cognitive load, cognitive fatigue, token spend, security; what changes for Scrum Masters and Agile Coaches; and finally governance, sovereignty and the gap between hype and reality. The book was itself drafted largely by AI, with every substantive claim checked against its source — a method it states up front and documents in a closing note.

What I took from it

• The role perspective, and the discussion of team size and of AI's impact on cross-functional (agile) teams.

Where it landed

• Nothing concrete yet. It sharpened my understanding.

Facilitator

Sohrab Salimi, Founder & CEO of Agile Academy — a fellow trainer and a mentor of mine.

Format

Training. Normally two days, in person or remote; for the ambassador cohort it ran as four afternoons.

Context

Attended as one of the Agile Academy’s ambassadors.

Material

The deck is licensed CC BY-NC-ND 4.0 — free to share, but not for public republication. Not reproduced here; characterised only.

The premise is that AI is infrastructure, not software: the ground the business runs on, touching People, Process and Product at once — which makes it a leadership question rather than an IT purchase, and explains why so few AI initiatives return anything. Its central move is the Player Coach: as the cost of delivering work collapses and the value-creating unit shrinks toward one person plus their agents, the coordinating manager role loses its reason to exist and leaders are pulled back into the work itself. The course design follows its own argument — leaders build an agent with their own hands on day one before being asked to lead anything on day two.

What I took from it

• Confirmed: AI is infrastructure, not IT. It has an effect on virtually every aspect and every step of the value chain of products, it can open new markets for new business models, and it reaches the people in the organisation and how they work together — including with AI — and therefore how people and companies are led.
• Confirmed: the delivery part of value creation gets much faster and more compact, and becomes possible with smaller teams, all the way down to “1 + AI” teams.
• Confirmed intuition: leadership moves back close to value creation, as far as actively shaping it — the Player Coach.
• Beyond confirmation: the Player Coach makes explicit what I had only thought intuitively — and why I am on this journey at all.
• New as an explicit point: companies and leaders should not outsource the topic of AI, because it has to become part of the DNA of the core business.
• New as a decision: how deeply to take AI into your own hands is an important choice — “only” the application layer, or the model layer, or even the infrastructure layer. In other words, what degree of sovereignty.

Where it landed

• Morten as an agent on Abundly, and Morten as part of a team. The training runs on Abundly too — Agile Academy is their first worldwide partner.
• Morten as part of one team — no dedicated marketing team any more, only a core team. “1 + AI” for the moment, maybe “2 + AI” later. Done on 31 July 2026: the GitHub team marketing became crew, Morten's role in it is Marketing & Sales, and the Slack channels were split into #crew for the normal case and #crew-alerts for escalation. Three days from the training insight to the org chart.
• The move to Player Coach was inspired here as well. It is datable: the line in my team profile about being “back in the game as a player-coach” went in on 28 July 2026 — the same afternoon the training covered the model, and before the source was named anywhere.
• I will check the learning journey for business-model viability — moonshot — the same way I check my existing business and the repositioning. And probably the tool we are learning on as well: the Chronicler.

Speaker

Nils Janse, Co-founder & Gen AI Expert at Abundly AI.

Format

Webinar, online, 59 minutes.

Material

The event page carries the description and the recording. Linked and characterised here, not reproduced.

The step that matters is not the single agent but the fleet: how agents across business functions are wired into end-to-end processes rather than left standing as isolated tools. It offers a repeatable way of getting there for any function — a method for turning scattered agent ideas into a structured portfolio — and shows concrete fleet examples. It takes an organisation that is already building agents as its starting point, and addresses executives and department heads, AI leads structuring an agent portfolio, and the champions and agent builders who want to see how their own agents fit the bigger picture.

What I took from it

• How demanding the road to a genuinely agentic organisation is.

Where it landed

• On the basis of this webinar I wanted to build the learning journey professionally, and to do everything that happens in a way that stays clean and maintainable even if I scale up to several agents.

Speaker

Nils Janse, Co-founder & Gen AI Expert at Abundly AI.

Format

Webinar, online, one hour.

Material

The event page carries the description and the recording. Linked and characterised here, not reproduced.

Governance is what makes agents scalable rather than what holds them back: who is allowed to create an agent, how agents get reviewed, what data and systems they may reach, how their performance is monitored, and who carries the risk. It works through the conversations to have with IT, the guardrails that let people build fast without giving up safety, and the ownership structures a growing agent portfolio needs. Addressed at AI leads, at the C-level accountable for direction, ROI, risk and security, at heads of IT and compliance, at engineering and platform leads, and at the department heads whose teams build and use agents.

What I took from it

• What matters to me is the triad of structure, alignment and ownership — and a governance model that sets the right frame at the right phase.
• I like the “Foundation – Factory – Fleet” steps as a way of growing organisations into agentic organisations.

Where it landed

• I worked in a considerably more organised and structured way afterwards: Morten's permissions, the PR gates, the separation of the repos.
• All of that is still the “Foundation” phase — “build your muscle” — but already with a view to building on it without tumbling into chaos.

Speaker

Henrik Kniberg, Co-founder & Chief Scientist at Abundly AI.

Format

Webinar, online, 80 minutes.

Material

The event page carries the description and the recording. Linked and characterised here, not reproduced.

How far agents can be taken in a real development workflow, grounded in Abundly’s own large and complex codebase rather than a prototype — the claim being a tenfold productivity gain and a new platform version shipped every day. It works through what holds up in practice: helping an agent navigate a big codebase, keeping control of the architecture when agents write most of the code, avoiding technical debt, and balancing agent autonomy with human oversight. It then goes beyond coding — agents for backlog and release management — and closes on what that means for team structure, processes and the engineering craft.

What I took from it

• The most important thing: it showed how easy the entry into agentic engineering is. Even more so with experience in software development.

Where it landed

• Direct connection: the journey started the same day, 17 June 2026, with the repo split, the characterisation tests and the Maven wrapper.
• Watching it was also a way to prepare myself a little for the “AI for Leaders” training that followed in July 2026.

Author

Henrik Kniberg — Certified Scrum Trainer with the Scrum Alliance, so a fellow CST; co-founder and Chief Scientist at Abundly, the platform Morten runs on.

Format

Video, 18 minutes.

Material

Fully public: the video itself, the complete drawing as a PNG on blog.crisp.se, a book on Leanpub, and AI voice translations into eight languages, German among them. Linked and characterised here, not reproduced.

A full day of AI course compressed into eighteen minutes of drawing and talking, aimed at everyone rather than at engineers. It starts from what a language model actually does — predict the next piece of text — and works outward from there to prompting, autonomous agents, the limitations that follow from the prediction mechanism, the origins of ChatGPT and AI-powered product development. It closes on the role of humans and on mindset: the tool has to be played with to be understood.

What I took from it

• The video put my understanding of AI at the time on a solid basis and sparked a deeper interest. Prompting on ChatGPT was already standard for me by then.
• Kniberg is a trusted source for me. He is a fellow Certified Scrum Trainer, and I already liked his “Agile Product Ownership in a Nutshell” and “Spotify Engineering Culture” videos and reference them in trainings.

Where it landed

• The video itself was not yet a driver to change anything. It created deeper understanding.