Heard this story a couple times now: Big consulting firm gets hired to "improve efficiency." They roll out textbook Scrum. Two-week sprints, story points, daily standups, the works. They collect their fee and leave. Six months later, employees are drowning in process overhead but efficiency hasn't budged. The problem: These weren't software companies. One was a research-driven hardware firm. The other, a large heavy-industry behemoth. At the hardware company, a researcher told me: "I can't even read a paper anymore. I have to write a user story about reading it, estimate how long it'll take, then report my progress every morning at standup." What's agile about that!? The industrial company? "Our work happens at massive scale with complex dependencies. Nothing here fits into two-week increments or neat user-stories." Now I'm watching companies make the exact same mistake with AI. Top-down mandates. One-size-fits-all playbooks from the same consulting firms. Going wide before going deep. Remember Gall's Law: Complex systems that work evolved from simple systems that worked. You can't jump straight to complex. Whether you're adopting agile or integrating AI, the approach is the same. Start with a small pilot. Learn what actually works for your organization. Build from there. Off-the-shelf transformations rarely transform anything except your budget. --- 📧 Daily newsletter on building AI/software projects that deliver. Link in profile.
Brian Sutherland
5mo
Your closing remark really says it all... Off-the-shelf transformations rarely transform anything... What really resonates with me is your callout to Gall's Law. We've grown too accustomed to being the first to the finish line, but in doing so we fail to notice the smaller opportunities that, when added up, result in greater value than what we were racing to.
Allen Smith
5mo
The parallel between forced Agile rollouts and rushed AI adoption is spot on. Real transformation comes from experimentation and iteration, not copy-pasting frameworks that ignore context. Gall’s Law should be required reading for every exec driving AI efficiency.
The only good thing about scrum are the agile short feedback cycles. Everything else is just process over people, which is exactly the opposite of the agile manifesto. I've never seen scrum work well in any context and people being happy about it. People just go along with any inane BS and focus on getting their job done fighting whatever system is trendy at the moment. Before, during waterfall, they were wasting time writing hyper detailed documentation (I see the value in documentation, especially after 15 years of agile, but back in the days it was just too much); now they waste time in useless meetings which should have been an async slack message. The best teams in corporate-ville are the ones were people know to do the bare minimum to check the ritual box and keep management happy and then focus on what's really impactful. Top down mandates in general are a dumb idea: management doesn't have visibility on what's needed and forcing everybody to follow the same playbook means you either get it right or your company will be in trouble. Decentralising power - and being generous with rewards and ruthless with failure - will make sure only the teams who made the right choices and achieved the best results will survive.
Artur Sossin
5mo
AI has so many business leaders in heavy FOMO mode. It's mind-boggling to say the least. Starting small hedges your investment and keeps the downside low, plus you start building the capabilities and testing culture fit in the org. You can always scale gradually from there. PS. Clemens Adolphs, I used to work in research, pure scrum brought us nothing but pain and inefficiency ...
Sunny A.
5mo
Agile and AI both fail the same way when they’re treated as plug and play solutions rather than adaptive systems. Real transformation is contextual, not templated. The best outcomes come when teams start small, learn fast, and evolve practices that fit their actual work, not someone else’s slide deck.
Colm BYRNE
5mo
scrum is a framework, not a process like The US Constitution is a framework, not a process if you set up a new country with the existing framework you will get yourself an entirely different country to the US Similarly, a country with the US constitution can turn authoritarian by hollowing it out like loyalists in the courts, administration, neutralizing checks and balances. Same with Scrum. It can be implemented in such a way that contravenes what the objective was in the first place!
Robert P.
5mo
At my last company, they implemented KPIs to improve performance. It was a flop, since the rollout occurred alongside major system updates. It's hard to plan or focus on your KPIs when you work for a startup and the ground is shifting from week to week. I think KPIs are a valuable strategy for dinosaur companies where they already built the moat and they are trying to squeeze out maximum productivity. For startups, it seems like an exercise in futility. There were other failed initiatives like that, the end result? The company is now bankrupt because they wasted time on futile change management efforts rather than focusing on customers.
I've heard this kind of story many times, and it always misses the same key points. a) What those consultancies roll out is not textbook Scrum. Scrum doesn’t mandate sprint length, story points (those come from XP), estimation, or even user stories. It’s a lightweight framework for organising work: define long-term goals, set short-term iteration goals, plan, align daily, review with stakeholders, reflect, adapt, repeat. That’s it. It’s a cadence of inspection and adaptation — something every healthy organisation needs anyway. b) The fact those companies weren’t software firms is irrelevant. Scrum isn’t about software. It’s about managing complexity and uncertainty. You could use it to paint a house if you wanted to. c) Writing a user story for reading a paper is not Scrum. It’s cargo cult. It’s what happens when consultancies sell templates instead of understanding. 1/2