Vijay Raina has spent years navigating the high-stakes world of enterprise SaaS, where the friction between dreaming of a perfect user experience and the cold reality of engineering constraints is a daily battle. As a specialist in software architecture and design leadership, he brings a grounded perspective to the often hyperbolic conversation around artificial intelligence. Rather than focusing solely on whether AI will steal jobs, Vijay examines how these tools are dismantling the old “permission culture” of tech companies, forcing designers to step out from behind their Figma prototypes and into the messy, results-oriented world of product ownership.
This conversation explores the diverging paths for digital design: one where designers gain unprecedented agency to ship improvements directly, and another where they are sidelined by “plausible” AI-generated mediocrity. We delve into the end of excuses for design debt, the potential for massive shifts in team sizes, and the new requirements for designers to possess commercial judgment and technical curiosity.
Designers often struggle in the “middle space” where product defines problems and engineering determines feasibility. How has this traditional structure limited the impact of design, and why is that changing now?
For a long time, designers have been trapped in a reactive loop where they are expected to make things “usable” or “desirable” without actually touching the machinery that makes the product work. You see it in every design review: the designer identifies a broken onboarding flow or a confusing upgrade path that essentially makes half of your new users feel lost, but they are told to wait. Because they don’t own the production tools, their ideas are just suggestions that have to be negotiated into a roadmap that is already packed with other priorities. This position is incredibly uncomfortable because you have the taste to see what is broken, like an empty state that makes a user feel stupid, but you lack the direct power to fix it. AI is fundamentally shifting this by moving the boundary of what a designer can actually build, meaning they no longer have to wait three months for a roadmap slot just to clean up a piece of design debt.
You’ve mentioned that AI allows designers to operate with “less permission.” What does it look like when a designer moves from being an internal critic to someone who can push a fix live?
It changes the entire politics of the office when a designer can move from saying “we should fix this” to “I fixed this, and here is the working version.” In the past, design was an argument supported by Figma prototypes and research clips, but now, a motivated designer can use AI to bridge the gap between a vague idea and a tangible interaction. You might see a small edge case that is actually ruining the first-run experience for a huge chunk of your audience and, instead of begging for engineering time, you can prototype the alternative and write the product copy yourself. It makes the “better thing” so visible and functional that it becomes almost impossible for leadership to ignore. This autonomy turns a designer into a hybrid product leader who isn’t just providing “taste” but is actively making trade-offs and resolving messy implementation details on the fly.
While the “bull case” for AI is about empowerment, the “bear case” suggests that autonomy might expose the gaps in a designer’s skill set. How will this new transparency affect those who are used to hiding behind organizational constraints?
Autonomy has teeth, and it is going to be very unkind to designers who have spent years using “lack of engineering time” as a shield for mediocre thinking. For a long time, it was easy to stay in a safe position of opposition, claiming you had a better idea that simply wasn’t feasible to build. But if AI gives you the tools to make that alternative real, then your idea has to actually survive contact with reality and solve the problem better than the existing version. We are going to find out that some designers aren’t as strategic as they thought; they’ve learned the vocabulary of “user needs” and “systems thinking” without ever having to own the actual outcome. When you can fix the debt yourself, you have to decide if it was actually worth fixing, and that level of exposure is going to be a wake-up call for the profession.
There is a significant concern regarding “plausible design”—the idea that AI can generate work that looks good enough to pass a review but lacks depth. How dangerous is this for the future of product quality?
Plausible design is perhaps the most dangerous trend we face because it creates a world of coherent mediocrity that nobody feels strongly enough to object to. An AI can produce a flow where the spacing is fine, the components are used correctly, and the copy isn’t embarrassing, which allows a Product Manager or an Engineer to bypass the design team entirely. In a product review, these designs look “good from afar,” and since many companies don’t actually know the difference between great design and something that is just “plausible,” they ship it. This leads to products that feel fine on the surface but are quietly damaging the business because they lack the deep interaction thinking and judgment that a human expert provides. If design is reduced to just maintaining the furniture and policing component usage, the actual shaping of the product will be handled by people who prioritize speed over substance.
We often hear that AI will automate the “boring 20%” of a designer’s job, but you’ve suggested the impact on team sizes could be much more drastic. Why do you see a 50% reduction as a more likely scenario for large organizations?
The “boring 20%” argument is a comfort blanket that ignores how large tech organizations are actually structured. Many design teams grew to their current size not just to do “design,” but to handle the heavy coordination, specialists’ handoffs, and the slow production cycles inherent in old-school software development. If AI reduces that scarcity of production time and makes prototyping nearly instant, you simply don’t need the same headcount to maintain the machinery of the process. In companies where leadership never truly understood why the design team was so large in the first place, the cuts could easily reach 50% or more as they realize a smaller, more capable group of hybrid designers can do the same work. The future likely holds fewer designers overall, but those who remain will have significantly more direct influence over the core product than they ever did in a bloated department.
For a designer looking to thrive in this shifting landscape, what specific skills should they prioritize beyond their traditional craft?
Taste is no longer enough; you need to pair it with technical curiosity and a sharp sense of commercial awareness. The designers who succeed will be the ones who can move seamlessly between a customer conversation, a pricing concern, and a messy implementation detail without feeling like those tasks “belong to someone else.” You need to be comfortable prototyping in code, or at least close enough to it that your designs aren’t just pictures, but functional logic. It’s about having the nerve to make decisions before every single variable is settled and the judgment to know which of the dozens of AI-generated options is actually worth pursuing. You have to become a maker who understands the business shape of the problem, rather than just a specialist who focuses on the elegance of the argument.
What is your forecast for the digital design profession over the next few years?
I believe we are heading toward an uncomfortable mix of both the bull and bear cases, where the gap between the best and the average becomes a canyon. We will see a small elite of designers who act as true product partners, using AI to move with incredible speed and shipping high-quality work with almost no friction from the organization. At the same time, many large companies will settle for plausible mediocrity, using AI to generate “good enough” interfaces and pushing their remaining design staff into roles that look more like governance and maintenance. The profession is being tested; for years we’ve asked for more power and less constraint, and now that AI is providing exactly that, we are going to see who can actually deliver value when the excuses are stripped away. Some will prove they were the strategic leaders they claimed to be, while others will realize the constraints were the only things making them look necessary.
