There is a specific kind of expensive mistake that organizations make repeatedly, across every domain of their operations. It happens when a representation of a system—a map, a plan, a model, an architecture—becomes so accepted that it stops being treated as a representation and starts being treated as the thing itself. Decisions get made. Resources get committed. Transformations get launched. And the system, indifferent to the confidence projected onto it, behaves according to its own actual structure rather than the one that was agreed upon in a workshop or a planning cycle. 

The damage isn’t caused by ignorance. It’s caused by the specific confidence that comes just after you’ve done the work to understand something. You’ve run the value stream mapping exercise. You’ve built the quarterly roadmap. You’ve deployed the AI agents. You’ve optimized the workflow. The effort creates a feeling of clarity that’s hard to question—because you did the work to earn it. 

Taz Brown has spent years watching this dynamic unfold in value stream transformations. The maps look clean. The swim lanes are documented. The team nodded. And the map is wrong—not because anyone was careless, but because the exercise was built on assumptions nobody challenged. Her practice of Red Team Thinking—an adversarial discipline drawn from military and national security contexts—is a structured antidote to this. “Flow practitioners are trained to see waste,” she says. “Red Team Thinking trains you to see the waste in your thinking.” The question isn’t whether you mapped. It’s whether you ever seriously asked what you assumed to be true while mapping, and what would happen if those assumptions were wrong. 

Ester Overjero, CEO of Vecta Global, extends the same problem into a different dimension. Even when the map is accurate, it may still be drawing a boundary around the wrong unit of analysis. Her systemic transformation work has surfaced a consistent pattern: local optimization—improving a workflow, accelerating a process, leaning out a team — regularly produces ripple effects in the broader system that weren’t visible from inside the thing being optimized. Faster here creates fragility there. Better in one stream degrades performance in another. “The real breakthroughs don’t come from making workflows faster, leaner, or more efficient in isolation,” she says. “They come from understanding the system as a whole.” The problem isn’t just that the map is wrong. It’s that the map may be drawing the right picture of the wrong territory. 

Stephen Walters, Field CTO at GitLab, sees the same failure reproducing itself in the current wave of enterprise AI adoption. Organizations are acquiring the technology and deploying it—but the hard thinking about coherent architecture, shared context models, and well-designed handoffs is being deferred. The result is AI agents operating in exactly the same fragmented, siloed conditions that produce poor outcomes when human teams operate in them. The technology takes the blame. The actual cause is structural. “The success of your AI agents depends not on the sophistication of the LLMs they use, but on the clarity and coherence of the context in which they are allowed to operate,” Walters says. The capability was real. The system it was dropped into was incoherent. The confident deployment obscured the absence of the thing that actually determines whether it works. 

Colleen Johnson, CEO of ProKanban.org, identifies the same mechanism in the way organizations plan. The quarterly roadmap is, at its core, a device for converting uncertainty into the appearance of certainty. Teams commit months in advance based on assumptions that are already outdated before the quarter begins—and because those commitments feel like plans, they resist the adaptation that actual conditions demand. WIP balloons. Context switching degrades quality. Everyone is busy; nothing gets finished, and the planning cycle that was supposed to create control has instead deferred the cost of its own inaccuracy until it’s compounded. “The assumption that work can be planned, estimated, and sequenced accurately for an entire quarter is fundamentally flawed,” Johnson says. Her model of continuous portfolio flow replaces the batch plan with a system in which work enters only when there is demonstrated capacity to start it—smaller decisions, fresher data, and the ability to course-correct before the gap between the plan and reality becomes a chasm. 

The pattern across all four is precise: An artifact designed to create alignment—a value stream map, a workflow optimization, an AI deployment, a quarterly roadmap—gets treated as more reliable than the system it represents. The confidence that artifact generates becomes an obstacle to honest interrogation. And the cost of that confidence shows up later, in transformation initiatives built on flawed foundations, in AI systems that contradict themselves, in portfolios where everything is planned, and nothing flows, in workflows that look efficient and quietly weaken the system around them. 

The antidote isn’t less rigor. It’s a specific kind of second-order rigor—turning the same analytical attention onto your models that you normally direct at your operations. Adversarial questioning of your current state maps. Systemic perspective on your local improvements. Contextual coherence in your agentic architecture. Capacity-based decisions in your portfolio. 

Taz Brown, Stephen Walters, Colleen Johnson, and Ester Overjero are all speaking at Flowtopia Live—and together, they make the case that the most important capability in transformation work right now isn’t optimization. It’s the willingness to honestly interrogate the systems you’ve already convinced yourself you understand. 

Join us on June 24th and get flowing. 

Exclusive Offer: As part of the TechStrong community, get a free LIVE pass. 

Use Code: TECHSTRONGFLOW 

Get your Live Pass to Flowtopia Live 2026 here.