The Rules are the Art: How Talos Design is Raising the Ceiling | Notes from Figma Config 2026

Introduction

AI has made it easy to be average. At Config 2026, Figma’s annual conference where the entire Talos Design team recently convened, the design industry’s response to this got a name: “raising the ceiling”. 

Dylan Field, Figma's CEO, opened with the framing that shaped the rest of the week: AI has “lowered the floor” (the barrier to entry for design) but it hasn't raised the ceiling. Designers now have to do that. While this was part of a Figma pitch, the observation still resonates: once producing something passable becomes easy, the only thing left to differentiate a design team is the value it adds above that baseline.

At Talos, where we design the platform that institutions rely on to trade and manage digital assets across the full investment lifecycle, that value added is trust, which comes from the human judgment AI can’t automate. 

‍

‍

Why “good enough” is a risk for Talos users

AI lowering the floor is good news on its own. It removes drudgery and gets any team to a working baseline fast. The risk is that once everyone can reach the average baseline, differentiation disappears. For Talos users running complex trading workflows at institutional scale, using execution algorithms across an ecosystem of more than 100 integrated providers, “average” introduces risk. Any friction in the user experience at the wrong instant can incur real cost.

In that setting, if a trader has to pause mid-trade to confirm an order actually filled, or to work out which screen shows the live position, that hesitation is a failure of the interface design at the moment it matters most. A user's experience is a portion of someone’s life they’re never getting back, and for Talos users, that happens mid-trade, when confidence matters most. Good user interface design either engenders confidence in that moment or it quietly costs the trader time and certainty. 

Holly Herndon, an artist who has spent years building work where the system itself is the art, offered a useful way to hold onto that standard. She described two ways of working with AI: as a centaur, a human orchestrating AI systems, or as a minotaur, a human orchestrated by them. Same tools, opposite dynamic. The question is who is actually making the calls.

Staying in the centaur role, directing the AI rather than accepting its first output, is what keeps "good enough" from becoming the standard. It's the difference between shipping the first plausible output and the one a designer has actually opined on and refined: checked for clarity, tested under pressure, and stripped of anything that adds friction to a workflow. The resulting trust in the product fosters the user’s confidence to move through a complex trading workflow without double-checking every step. Trust in the brand fosters the confidence to engage with Talos at all.

‍

‍

The design system, not the tool list, is where the craft now lives

Herndon gave the practice a name: protocol art. When AI handles execution, the rules of creation become the work itself — the site of production moves up one level of abstraction. You're no longer crafting the artifact; you're crafting the system that generates it.

This reframed something we'd been mulling for a while. We went to Config half-expecting a shopping list of tools. We left convinced the tools were the second question. The first question is the protocol — what should the workflow be, and what could productivity gains unlock: not only time for new features, but also more research, more stability, more confidence in each decision. As captured in one of the conference sessions: “good questions beat good answers.”

For Talos, the “rules of creation” are concrete: the design system, the component patterns, the intelligence and context layers that power brand experiences, the shared guidance handed to AI agents, and the reusable checklists that ensure nobody is starting from scratch on a new screen, campaign, or asset. One line from a Config talk stuck with us: “you can't prompt what you can't name.” A design system that's well organized, well linked and searchable is the medium where the work happens now. It’s where the craft lives.

Shrinking the loop between design and build

The hardest part of design isn't the idea. It's getting that idea’s intent to survive all the way to what a user actually sees on screen. Figma shipped real tooling for that this year: code layers that turn a design into editable code on the canvas, GitHub sync, and motion and shader support that bring implementation up into the design file. The loop between designing something and shipping it is shrinking fast.

That's the direction Talos is building toward. The team's hardest problems right now are keeping the design system and its code implementation in sync, and making consistency something the system enforces automatically rather than something we catch by hand. Naming those problems precisely is what lets the team address them, and it's the work we're most energized by.

In her talk Design System Anarchy, Lauren LoPrete offered the metaphor we keep coming back to: “a design system is like a legendary music venue.” Its job isn't to control the bands; it's to create the conditions where great performances happen. That's how we think about governance: not a rulebook, but a stage that lets the team play at its best while protecting design intent.

‍

‍

The audience's response is the final step

Brent David Freaney made a related point in his talk, Death of the Finished State: “a design isn't finished when it ships. It's finished when it meets a person.” The audience's response to design is the final step in the design process itself. For Talos, that person is placing a trade, checking a settlement, and deciding whether this is the platform they can rely on under pressure.

AI made average easy to achieve. What's additive is judgment, taste and the discipline to earn a trader's trust and confidence. This is the actual work. This is the ceiling we intend to keep raising.