Navigating the AI frontier: Projects across media

For twenty-five years, the same paradigm has governed knowledge work: your content goes to the tool. You reshape your thinking to fit Notion’s structure, Confluence’s data model, Jira’s workflow. But a shift that began with responsive web design in 2000 has been quietly unbinding content from its containers — first the screen, then the device, then the tool ecosystem, and now the application itself. This post traces the full arc, proposes a practical framework for the inversion, and asks the question nobody’s addressing: when intelligence is everywhere, what keeps us thinking for ourselves?

When a technology shift is small, existing mental models stretch to accommodate it. When the shift is categorical, everything cracks – and for the first time, the turbulence is strong enough to trace exactly where the cracks run. This exploration follows the loop between individual mental models and organisational structures that keeps the AI-as-tool paradigm locked in place, introduces a dual-perspective diagnostic that accounts for both your literacy level and your environment’s maturity, and argues that the missing dimension – metacognitive flexibility – is what no current framework measures and what the shift actually demands.

Over the past year, I’ve been exploring how AI reshapes human cognition — mapping cognitive offloading, trust paradoxes, automation bias, and the shift from click-based to agentic interfaces across the Stimulus blog. In early 2026, two comprehensive research compilations landed on my desk, synthesising the latest peer-reviewed studies from MIT, Microsoft Research, BCG, KPMG, and the World Economic Forum. The convergence was striking: seventeen points where the science now validates what we’d been exploring. This post maps those connections — not as a victory lap, but as evidence that following good science forward with genuine curiosity leads somewhere real.

What happens when you point the Orchestration Load lens at UX practice itself? The diagnosis is uncomfortable: the system turned our methods into rituals. Sprint pressure, process ceremonies, and production overhead consume 85-90% of a designer’s time — leaving the strategic work the discipline exists for squeezed into the margins. But AI is collapsing the production layer, and 25-30 hours of cognitive capacity are about to open up. The question isn’t whether UX can make this transition. It’s whether we’ll fill the freed capacity with the strategic work we were always meant for — or find new overhead to consume it.

What happens when a perfectly designed AI tool makes its users quietly worse at their jobs — and nobody notices? The Orchestration Load Framework introduces six types of cognitive load that humans carry when working with AI, revealing why traditional UX metrics miss the most consequential failures. Through independent audits of 10 AI tools, one finding stands out: how you present AI output matters more than how good the AI is. This is a UX problem — and UX practitioners are the right people to solve it.

Cognitive biases aren’t bugs—they’re features. Until they become rigid, exploited, imbalanced, or stop adapting. Discover the four-layer framework mapping how we think individually, collectively, under manipulation, and when merged with AI. Your cognitive sovereignty depends on it.

Here’s an uncomfortable truth: most “user-centered” design teams are building features users don’t actually want. Not because the features are poorly designed, but because users don’t want features at all-they want outcomes. Using the inversion thinking framework, we discover that Design Thinking and Agile Development are broken in exactly the same way: both have become feature factories disguised as outcome-driven processes. Once you see it, you can’t unsee it.