Winjit Launches “Beyond Screens,” a research paper rethinking service design for the agentic AI era, at GFF 2026

What happened

Winjit has released a research paper titled "Beyond Screens," presented at GFF 2026, that argues service design principles need to be rethought for a world in which agentic AI systems act on their own, with minimal human oversight. The paper's central premise is that as AI agents move from answering prompts to independently executing multi-step tasks, the conventional model of designing interfaces and journeys around human clicks and screens no longer holds.

According to the reporting, the paper positions this shift as a structural one for the design profession: when software can initiate, decide and complete actions without a person navigating each step, the unit of design shifts away from the screen itself towards the rules, guardrails and outcomes that govern autonomous behaviour.

Why it matters

This is fundamentally a story about how agentic AI changes the practice of designing services and products, not simply a customer-experience footnote. If AI agents are increasingly the ones interacting with systems on a person's behalf, then traditional service-design tools — wireframes, journey maps, screen flows — address a shrinking part of the actual experience. Organisations that keep optimising visible interfaces while agents transact invisibly in the background risk designing for the wrong layer entirely.

For leaders building or buying AI capability, the paper's framing suggests a practical shift in where design effort belongs: towards defining what autonomous agents are permitted to do, how they escalate uncertainty, and how outcomes are made legible to the humans ultimately accountable for them. That reframing has implications well beyond any single vendor or sector, touching any organisation planning to deploy agentic AI in customer-facing or operational roles.

The Renascence take

Publications like this tend to arrive well ahead of the operational reality they describe, and it is tempting to file "rethink design for agents" alongside the last decade of "rethink design for mobile" or "for voice." The more interesting claim buried in this one is behavioural, not visual: when the user of a service is another piece of software rather than a person, the trust-building mechanisms designers have relied on — reassurance copy, progress indicators, confirmation screens — stop doing their job, because there is no one there to reassure.

The real design problem in agentic AI isn't the absence of a screen — it's the absence of a moment where a human can sanity-check what's about to happen. Most organisations will respond by trying to bolt dashboards onto agent activity after the fact, which just recreates the screen they thought they'd eliminated. The operators who get this right will instead design the exception paths first: the specific junctures where an agent must pause, disclose its reasoning, or hand control back to a person, and treat everything else as genuinely automatable. That is a governance and behavioural-design exercise as much as a technical one, and it belongs on the same roadmap as the AI deployment itself, not bolted on afterwards.