In 2024, I joined the Online Scrum Master Summit for the first time, an event filled to the brim with insights that inspired me as an agile coach and scrum master. The summit has a broad roster of speakers from across the world, each bringing their own view and practical guidance. This year marks its fifth anniversary (congratulations!), and as I did with the last edition I joined, I want to cover the topics that struck me most in a few blog posts. This one gathers my personal experiences and key takeaways from the first day, Wednesday, June 17th. There are four talks I want to highlight, and what stayed with me, watching them back to back, was how often four very different speakers kept circling the same quiet idea.
Interested in the program? Click here. You can also buy an all-access pass.
Maarten Dalmijn – Silosaurs Will Die
Maarten Dalmijn did something I always admire in a talk. He took the question everyone is asking, “what will AI change“, and turned it on its head. Ask instead what won’t change. His answer was that “silosaurs will die”. These are the organizations that keep their people in cages, where the process matters more than the point, and then act surprised when nobody flies anywhere together. Strapping AI wings onto a caged bird, as he put it, misses the obvious. The cage was never about speed.
What stayed with me was his argument that trust, not AI, is the thing actually holding most companies back. When people aren’t allowed to decide, giving them sharper tools just hands them competence they can’t use. Fixing trust, he said, has a higher return than fixing AI strategy. For anyone who has spent years arguing that psychological safety is structure and not sentiment, this landed as quiet vindication, dressed in welcome fresh language. You can’t be the human in the loop, he reminded us, unless you’re genuinely smarter than the thing you’re correcting.
He closed on roadmaps. Show him your roadmap and he’ll tell you your problems, because every conflict you’d rather avoid eventually surfaces there. Red roadmaps railroad. Blue roadmaps leave room to play. I won’t spoil the rest, partly because the penguins deserve to be experienced live, but if you were in the room you already know why he was a favourite last year.
Aino Vonge Corry – Retrospectives Antipatterns
Aino Vonge Corry came to OSMS26 with a premise I wish more speakers were brave enough to admit out loud: she has made the mistakes herself. Her whole approach to retrospectives is built on antipatterns, which as she puts it are like patterns, only more informative. You first recognise the shape of a bad retrospective, then you learn how to climb out of it. The framing is generous and a little disarming. Nobody is being told they’re doing it wrong from on high. Everyone in the room has run the boring retro, the one that ran late, the one where the same complaint surfaced for the ninth time and nothing changed. She just happens to have names for all of them.
What I appreciated most was how firmly she anchored the craft in respect for people. She opened with Norm Kerth’s Prime Directive, the reminder that everyone did the best job they could with what they knew at the time, and treated it not as a poster on the wall but as a working tool you actively bring into the room when blame starts creeping in. There is nothing soft about that, whatever the word “care” suggests. Done well, care is a working discipline, the thing a good retrospective is actually built on. Her antipatterns, the retrospective nobody prepares for, the facilitator who quietly does everything themselves, the team stuck stewing in problems they can’t actually influence, are all really failures of care dressed up as process problems. Name them honestly and they lose most of their power.
I won’t catalogue them here, because the octopus illustrations and the war stories are half the reason to watch the recording. But the through-line is worth carrying into your next session of any kind: a retrospective only works when people leave feeling heard and a little more able to change their world than when they walked in. That holds whether you’re facilitating a sprint retro or any other meeting where people are asked to be honest with each other.
Konstantine Kevlishvili – Rediscover Scrum Basics: Are You Delivering Too Much?
Konstantine Kevlishvili did the bravest thing a speaker can do: he started with the mirror. He’d been reading the product canon, Cagan, Perri, Torres, and recognising the symptoms everywhere. Building things with no real value, measuring output over outcomes, becoming a feature factory, skipping discovery entirely. Then he looked up and realised the books weren’t describing some other company. It was us. Coming from someone coaching agile teams at a FTSE-250 bank, that admission set the tone for everything that followed. This wasn’t a lecture about other people’s failures. It was a confession, and a generous one.
His diagnosis is one I suspect a lot of us recognise. Good teams fall into the delivery trap not through incompetence but because output is easy to count and shipping feels productive, while discovery feels like falling behind. He reached for Goodhart’s Law and the old colonial-Delhi cobra bounty to make the point stick: reward the number and you get the number, not the outcome. And his warning about AI was sharper than most. If delivery becomes almost free, we just ship more, get overexcited, and hit the same wall faster. The cure, he argued, isn’t a new framework or a clever tool. It’s older than the disease. It’s Scrum read properly, as a learning engine built on transparency, inspection and adaptation, rather than a delivery conveyor belt.
What made it land was that he’d actually tried it, with two pilot teams at the bank, and was honest about how hard it was. Getting people into a room with a real customer every week, then resisting the urge to ask leading questions, then falling out of love with ideas they’d already committed to. That last one, he said, is the real skill. I won’t give away his Monday-morning challenges, because they’re worth hearing in his own words, but one has stuck with me: stop asking “is it done?” in the Sprint Review and start asking “did it work?”
Dave West – AI Catalyst: The Scrum Master’s New Role
Dave West, CEO of Scrum.org, opened OSM 2026 with a paradox worth sitting with: AI is genuinely accelerating individuals, but those individual gains do not automatically become team gains, and nobody owns the gap. The evidence he stacked makes the point land. Certain tasks run 55% faster in controlled studies, yet GitClear’s longitudinal research found code duplication quadrupled as AI adoption rose, and a METR trial found no statistically significant productivity gain even as participants believed they were 20% faster. The bottleneck has moved off execution and onto coordination, judgment and design. That is not a new problem. It is the Scrum Master’s oldest one, wearing new clothes.
What makes the talk useful rather than alarming is the reframe. The Scrum Master as “AI Catalyst” is not the team’s AI expert but its AI effectiveness leader, and West is explicit that you do not need to be the most technically fluent person in the room to do the job. The work is to make the team behave as a system rather than a collection of individuals, through three moves: diagnose where the team sits on the fluency ladder and which engagement mode it is actually in; design by upgrading existing Scrum events instead of bolting AI ceremonies on top, adding quality gates to the Definition of Done and protecting the Sprint Goal from five accelerated developers building five diverging solutions; and develop shared team assets while guarding against “capability debt,” the quiet erosion that happens when AI absorbs the execution work that used to build junior practitioners’ judgment.
The line that stays with me is his closing one: the role becomes more human as the team becomes more hybrid. As AI tempts people toward isolation and individual speed, the differentiating work is psychological safety, facilitation, critical thinking and governance. That is the “Human Edge” argument arriving from the CEO of Scrum.org, in his own vocabulary. It is also, quietly, the case for treating the so-called soft work as the hard work, the structural discipline that turns five fast individuals back into one capable team.
Four speakers, four very different talks. A caged bird, an octopus, a cobra bounty, a fluency ladder. The same shape keeps surfacing. Dalmijn calls it trust, Corry calls it care, Kevlishvili calls it the discipline of falling out of love with your own ideas, and West calls it the “Human Edge”. They are all pointing at the work we keep filing under “soft,” the work that turns out to be the structure everything fast and clever has to stand on. AI did not invent that work. It only raised the stakes on getting it right. Which leaves me with the one question I have been turning over since the first day closed: if the soft work really is the hard work, what would it take for our teams to start treating it that way?
Stay tuned for my next blog post, in which I will discuss the highlights of the summit’s final day and share additional key takeaways.