Config 2025 was interesting because Figma stopped looking like a tool that ends at design. Make, Sites, Draw, Buzz, Grid improvements, and deeper developer workflows all pointed in the same direction: the distance between an idea, a designed system, and something people can actually use is getting shorter.
AI was everywhere at Config 2025, but focusing only on AI misses the more consequential change. Figma was expanding across prototyping, code-adjacent creation, websites, illustration, brand production, and developer collaboration.
For product designers, that changes the shape of the work. A screen is increasingly just one representation of a system that also has components, states, content rules, responsive behaviour, accessibility requirements, technical constraints, and implementation consequences.
The opportunity is not simply to produce more screens faster. It is to carry intent further through the product-development cycle.
Prompt-to-working-interface tools make it dramatically cheaper to test an idea with behaviour instead of describing it with static frames. That is especially useful early, when a team needs to understand whether an interaction model makes sense before investing in polished production design.
But generated output is not automatically product thinking. A convincing prototype can still contain weak hierarchy, impossible states, inaccessible interactions, inconsistent components, or assumptions that collapse under real business rules.
I see Make-type workflows as a way to make an argument tangible faster. The designer still owns whether the argument is any good.
Sites made another long-running boundary less rigid: the point where design ends and a web implementation begins. For lightweight sites and communication experiences, moving from composition to responsive output inside the same environment can remove a surprising amount of translation.
That does not make engineering irrelevant. It makes technical literacy more valuable. The closer designers get to production, the more important it becomes to understand responsive behaviour, semantic structure, performance, accessibility, reusable patterns, and what should become a system rather than a one-off.
The less theatrical improvements are often the ones that matter most on enterprise teams. Better layout logic and stronger connections between design intent and implementation reduce ambiguity where it is expensive: responsive behaviour, component decisions, spacing, states, and the small inconsistencies that accumulate when design and code tell different stories.
This is why I increasingly think of a design system as a working contract between disciplines. The goal is not perfect parity between a Figma file and production. It is shared logic that survives both.
Draw pushes Figma further into expressive visual work; Buzz pushes it toward scaled brand production. They solve different problems, but together they reinforce a useful idea: design systems do not exist only for product designers.
Organizations need systems that can support product, marketing, content, brand, and other teams without turning every asset into a bespoke design request. The challenge is giving more people controlled flexibility without dissolving the quality and consistency the system was meant to protect.
Generative tools make creating alternatives cheap. That sounds liberating until every generated alternative introduces a slightly different button, spacing rule, content pattern, or interaction model.
Strong systems become guardrails for AI-augmented work. Components, tokens, interaction rules, accessibility expectations, content conventions, and known implementation constraints give rapid generation somewhere coherent to land.
Without that structure, AI can accelerate design debt as efficiently as it accelerates exploration.
As tools absorb more production work, the valuable parts of product design stretch in both directions. Upstream, designers need stronger problem framing, research synthesis, product judgment, and strategic clarity. Downstream, they need greater awareness of implementation, system behaviour, accessibility, and whether the shipped experience preserved the original intent.
That is a more demanding role, not a diminished one.
Config 2025 did not convince me that designers should become full-stack everything. It convinced me that the strongest product designers will understand more of the system around their work—and will take ownership of what happens to an idea after the polished frame is finished.
