AI Design Tools Are Changing Fast — But Who Controls

The tools are faster. The outputs are more polished. And somehow, the design decisions feel harder to own than ever before.

That tension sits at the centre of where digital design is heading in 2026 — not a question of whether to use AI, but how to build a working practice around it that doesn’t quietly hollow out the thing that made your design valuable in the first place.

AI Tooling Has a Hidden Cost That Isn’t Listed on the Pricing Page

Speckyboy’s Eric Karkovack put it plainly this week: AI tokens aren’t free, and the design industry burned through its naive phase faster than anyone expected. What started as a billing-model curiosity is now a genuine workflow discipline. Teams using AI for everything from component generation to copy iteration are discovering that unconstrained usage creates both cost exposure and quality drift — the two things no creative director wants to explain to a CFO.

The practical implication for design leads in Southeast Asia is sharper than it might seem in markets with lower SaaS sensitivity. When your token budget is finite, you need a decision framework for when AI involvement adds value versus when a senior designer’s 20-minute judgment call costs less and produces something more defensible. Monochrome data visualisation — surfaced as an editorial pick by UX Collective this week — is a useful case study: constraint-driven design choices that make data clearer don’t benefit from AI generation; they benefit from human editorial clarity. The tool serves the principle, not the other way around.

For teams managing design systems across Shopee storefronts, LINE campaign assets, and app onboarding flows simultaneously, that discipline matters. AI can accelerate asset production; it can’t replace the person who knows why a certain visual hierarchy converts in Bahasa Indonesia but not in Thai.

The Interface Itself Is a Design Problem AI Hasn’t Solved

Smashing Magazine’s Oleksii Hrzhehorzhevskyi raised something this week that deserves more airtime than it usually gets: almost every AI tool we interact with defaults to a text box. That’s not a neutral design choice — it’s a significant UX constraint baked into the infrastructure layer, and it shapes what users think to ask for.

For design teams, this matters in two directions. First, the AI tools you use to create design are themselves often poorly designed for creative workflows — they reward verbose prompt engineering over visual intuition. Second, and more commercially relevant for brands, the AI interfaces your customers interact with are almost all built on the same text-box convention, which means there’s genuine first-mover opportunity in rethinking that interaction model.

Grab’s superapp interface and Lazada’s in-app discovery experience both hint at what non-text-primary AI interaction could look like — visual search, gesture-based filtering, image-led prompting. The design teams building those features aren’t waiting for OpenAI to solve the interface problem for them. That’s the right instinct.

---

Convention-Breaking Design Is Still a Human Competitive Advantage

The collaboration between International Magic and Martine Rose — documented this week by It’s Nice That — is worth examining not because virtual fashion shows are a template anyone should copy, but because of what it demonstrates about design intent. International Magic has built a practice specifically around resisting digital conventions: live web platforms that behave unexpectedly, visual identities that disrupt rather than confirm expectations.

That posture is commercially valuable precisely because AI tooling optimises for the mean. If your entire design team is using the same foundational models with similar prompts, you’re converging on similar outputs. The brands that will differentiate visually over the next 18 months are the ones with a clear point of view that exists before the prompt is written — a design philosophy that constrains and directs AI rather than being shaped by it.

For Southeast Asian brands specifically, this cuts against a common pattern: designing to platform spec (Shopee’s grids, Meta’s safe zones, TikTok’s text placement) and calling it done. Platform compliance is a floor, not a ceiling. The most recognised visual identities in the region — think Tokopedia’s distinctive illustration system or AirAsia’s typographic boldness — made choices that were harder to execute within platform constraints, not easier.

The Portfolio Problem Is a Trust Problem

UX Collective flagged it this week without much ceremony: AI can fake your portfolio. This isn’t a designer anxiety piece — it’s a systemic trust issue with real commercial consequences. If clients and hiring managers can no longer reliably verify design capability from presented work, the signals they fall back on are references, process transparency, and evidence of decision-making rationale.

For design teams at agencies and in-house brands, this changes how design work should be documented. The artefact — the final visual — is becoming table stakes. What differentiates a credible design practice is the audit trail: the brief interpretation, the constraint identification, the iteration rationale, the measurement framework for whether the design achieved what it was supposed to achieve.

From a data perspective, this is actually good news. Teams that have been rigorous about connecting design decisions to conversion metrics, engagement rates, and revenue outcomes now have a natural credibility advantage. The work isn’t just beautiful — it’s legible. You can explain why it worked. That’s not something a generated portfolio can replicate.

Key Takeaways

• Establish token governance before AI sprawl does it for you — define which design tasks genuinely benefit from AI assistance and which ones just consume budget without improving output quality.
• Build your design philosophy documentation before your prompt library — teams with clear visual principles use AI as a production accelerator rather than a creative substitute.
• Start treating design decision rationale as a deliverable in its own right; in an AI-saturated market, documented human judgment is increasingly what clients are actually buying.

The deeper question for design leaders isn’t whether their teams are using AI — of course they are. It’s whether the design principles guiding that usage are robust enough to survive the next tool cycle. Because the tools will change again. The question is what stays constant when they do.

---

At grzzly, we work with brand and marketing teams across Southeast Asia on exactly this intersection — building design systems and visual strategies that hold up across platforms, languages, and tool environments without losing the commercial intent behind them. If your team is figuring out where human design judgment ends and AI tooling begins, that’s a conversation worth having. Let’s talk