Working Notes from the Edge of Practice

In the past few years, probably a majority of what we would consider our most interesting work — certainly where we're stretching our own strategic foresight practice — has been invisible. Clients consistently tell us our value isn't in the way we run well-known tools and methods, but in how we think about their questions and how we explore them. That seems like lost value. Not just for us, but for anyone trying to understand what foresight practice actually looks like from the inside of a working consultancy.

R&D is an attempt to change that.

What gets built here doesn't always reach a deliverable. A database structure developed for a scenario planning engagement. A simulation engine rebuilt for a different risk domain to test whether the architecture held. A prompt system for turning development threads into publishable pieces. A positioning framework contributed to a global forecasting consortium. A small foresight operation run almost entirely by an AI system, with minimal human editorial direction, as a mirror for examining what the practice actually does versus what it says it does. These things get built, tested, iterated, pinned for later, returned to months on. The thinking behind them has mostly stayed invisible until now. The work described here spans several years. The decision to document it openly is more recent.

One thread running through much of this work has been the testing of generative AI tools — not as a productivity layer but as a genuine research question. Where do they fit within a foresight research process and where don't they? What do they change about how we design simulations, structure research, build frameworks? What do they reveal when they get things wrong? We've been working through this across audio, generative media, game engines, and other formats we're actively testing at the outer edge of what foresight and futures design can do. The answers keep shifting as the tools themselves evolve, which is itself part of what we're tracking.

“The thinking behind them has mostly stayed invisible until now.”

Running alongside this is the work logs themselves. Much of what we build gets developed through extended working sessions — in AI tools, in notebooks, across conversations that span weeks or months. We've started treating those logs as primary material, scraping them for the moments when something is learned, prototyped, discarded, or refined. When a design decision gets made and then reversed. When a tool responds in a way that changes the direction of the work. When something built for one purpose turns out to be more useful somewhere else. R&D captures those turns close to when they happen, before the learning gets smoothed into a method or forgotten entirely.

The most pointed example of what this is for is APE — a small autonomous foresight operation run with minimal human editorial direction, designed to surface what a foresight agency actually does when its judgment, selection logic, and framing choices are removed from the equation. It's an adversarial model of the practice: what would it look like if the role were challenged, or replaced, or simply made to explain itself? Not whether AI can do foresight, but whether the practice can describe what it does that AI doesn't. That's not a comfortable question. We think it's a necessary one.

"Not whether AI can do foresight, but whether the practice can describe what it does that AI doesn't."

What separates Changeist from other foresight practitioners isn't accumulated capability. It's the questions we're willing to ask about the edges of our work, the tools we'll test before there's a clear use case, and the things we're willing to challenge — including our own assumptions about what the practice is for. That gap doesn't show up in deliverables. It shows up in how we think, how we build, and what we're willing to discard when something better appears.

R&D pieces are short, artifact-first — captured close to the work, before the thinking hardens into method.