Last week a friend came to talk through a startup he’s building — an AI-powered branding system. A company inputs its business direction and positioning, and the system generates a full visual identity: logo, typography, color palette, brand guidelines, plus downstream extensions. The whole suite.
For most companies that want to take branding seriously, the established path has always been: find an agency, go through rounds of communication, iterate, revise, and absorb a budget that’s hard to justify unless you’re already at scale. A lot of small and mid-sized teams simply can’t afford it. AI is finally making it possible to solve this at a fraction of the cost. The demand is real. The market is there.
But as the conversation went on, I noticed the center of gravity had shifted. We weren’t really talking about how to build the product anymore.
Like many AI startups right now, the hardest part is no longer whether the thing can be built. It’s whether the problem being solved will still be a problem in six months — or even in three.
For years, the default playbook has always been: find a clear need, validate the market, build an advantage, then deepen that advantage into a moat. Even in crowded spaces, if a team moves fast enough, goes deep enough, and executes well enough, there’s theoretically still room. But AI has introduced a different kind of dynamic. A problem, once it becomes obvious enough, gets seen simultaneously — by startups, by foundation model companies, by platforms, by tool builders, across industries and disciplines. Teams spend six months constructing something that suddenly shows up as a feature in someone else’s next platform update.
The most subtle part is that this doesn’t happen because anyone did poorly. Often the opposite. Many of these teams are sharp, fast, and ahead of the curve — they recognized the shift before most people did. But the AI era has a particular kind of cruelty: the moment a problem has been clearly defined, it also begins losing its ability to belong to any single team or individual for long, because everyone converges on it almost simultaneously.
This has become more visible with each passing year. A few years ago, the discourse was still “can AI do X?” That question itself is approaching irrelevance. The answer is obviously yes — and usually far beyond what anyone imagined.
Meanwhile, a different set of questions is surfacing. If everyone can build fast, if all the tools eventually converge, where does lasting value actually get established? Is it taste? Worldview? Cultural sensibility? Some harder-to-replicate form of judgment? None of these have settled answers yet.
This is where a lot of teams are right now. Because reality doesn’t offer much room to pause and rethink from the ground up. Projects are still shipping. Clients are still pushing forward. Teams still need to keep running. AI capabilities shift every month, the industry shifts, client expectations shift. Just keeping up with the rate of change is already exhausting. So the natural response is to keep optimizing what’s directly in front, make the process a little more efficient, the output a little faster, the system a little more stable. All of that matters.
But in many cases, what’s actually happening is that people are solving, with great efficiency, a problem that has already started to expire. And the expiration is hard to detect from inside. It may even keep growing for a while. The metrics might still look fine, the clients are still there, the projects still move. It’s just that something deeper has already begun to shift.
Lately I keep thinking the question of what’s actually worth continuing to invest in.
Most forms of external support don’t operate at this layer. Consulting firms help companies analyze direction. Agencies handle execution. AI advisors help teams deploy tools and workflows. But in most cases, the default premise is that the problem itself is already valid. And the predicament an increasing number of teams face is precisely here, not that the solution isn’t good enough, but that the problem itself has started to move.
This is the essence of what Actual Fiction is doing now. Before how to solve it, there’s a prior step: helping teams see the actual landscape they’re operating in — what still holds, what has started to loosen, which assumptions that used to be default have quietly changed. Recalibrating what to do based on those shifts, before getting into how to do it.
Over the past several years, whether the work was product, creative, experience, or building something from zero — I’ve found that it all comes back to the same center: understanding what is actually changing between people and systems. The AI wave has just brought more industries into that state simultaneously.
Maybe the real difficulty of the AI era was never that change moves too fast. It’s that many problems haven’t been seriously understood before they’ve already begun to expire.
If any of this sounds familiar, let’s talk.
About the author
Qinqin (Stella) Yang is a director, builder, and imaginist working across AI capability and creative practice. She builds workflows, systems, and creative direction that operate under real constraints, and designs speculative frameworks that expand what teams can imagine before they start making. Over the past decade she has built brand experiences, product design, and speculative projects for Tencent, IKEA SPACE10, Burberry, and the Palace Museum. She is the founder of Actual Fiction, where she develops AI-native tools that turn imagination into a working method.
This newsletter is where she thinks out loud about creativity, AI, and what it means to build toward a future that doesn’t exist yet.
For more details: qinqinyang.com