Winjit's "Beyond Screens" paper rethinks design for agentic AI

What happened

Technology and digital engineering firm Winjit has published a research paper titled "Beyond Screens," unveiled at GFF 2026, that examines how service design needs to evolve as businesses move from screen-based digital interfaces towards agentic AI systems that act autonomously on users' behalf.

The paper argues that conventional service design thinking, built around visible interfaces, forms and dashboards, is being tested by agentic AI, where software agents initiate tasks, make decisions and complete workflows with limited direct human input. Winjit frames the release as an attempt to give designers, product leaders and technology teams a starting framework for rethinking how services should be structured when the primary "user" of an interface may increasingly be another piece of software rather than a person at a screen.

Why it matters

The shift towards agentic AI changes the basic unit of design. When agents can browse, transact and resolve issues without a human clicking through each step, the traditional screen-and-click journey stops being the main design surface. Organisations building or buying agentic AI capabilities will need new ways to define trust, control, transparency and error-recovery in systems where the "user" is often invisible and the decision loop happens in milliseconds.

For leaders in experience and digital transformation, this signals a coming redesign challenge: service blueprints, consent mechanisms and escalation paths built for human-operated screens may not map cleanly onto agent-to-agent or agent-to-business interactions. Getting ahead of this now, before agentic AI is embedded at scale, is likely to be easier than retrofitting trust and accountability into systems after the fact.

The Renascence take

Most commentary on agentic AI focuses on efficiency gains — fewer clicks, faster resolution, lower cost to serve. What tends to get missed is that removing the screen doesn't remove the need for the human behind it to feel in control.

The real design problem in an agentic world isn't the interface disappearing — it's trust becoming invisible along with it. Every screen-based journey today carries implicit reassurance: a confirmation message, a progress bar, a human to escalate to. Strip that away and organisations risk optimising for speed while quietly eroding the sense of agency customers rely on to feel safe handing a task to a machine. A customer-obsessed operator shouldn't just redesign workflows for agents; it should design explicit "trust checkpoints" — moments where a human can see, pause or override what an agent is doing — before agentic AI is running unsupervised across critical journeys.