English | EN 
When the web shifts from informing to acting
When my colleague Karl Fridlycke summarized his thoughts on Clawdbot and the movement forming around it in his blog post on the Swedish Sogeti site, it opened a new conceptual space for me. His description of agents continuing to work long after we’ve logged off made me reflect through my lens as a UX designer, on what really happens when we begin to delegate rather than just prompt.
When the agent becomes a coworker, not a tool
Karl describes how automated AI agents can accept tasks, push them forward, and take initiative based on context, not because the technology is here, but because the demand for reduced digital friction has finally become actionable.
A simplified analogy: in the same way you can set rules for your inbox to automatically sort email, your entire workday can be augmented by systems that think and act. How far you take it is up to you. A project‑planning meeting ends and the system continues executing what was agreed. A new project within an established architecture requires no more than clicking the “OK button”.
AI systems are moving from reactive co-pilots to autonomous actors that not only support work but exert real agency within digital environments.
From internet to autonet: when the web starts doing the work itself
What truly interests me is how this will transform our relationship with the web. For three decades, the browser has been the primary problem‑solving tool. But what happens when the system; your computer, phone, or AI agent can navigate, click, purchase, book, and compile on your behalf? Not because you explicitly asked it to, but because you needed it. It already knew it would help you accomplish your bigger goals.
Many now describe a digital future where intelligent automation layers and AI‑driven ecosystems blur the line between what we create and what systems silently construct. The web becomes an operational landscape, not an information destination. In that light, “autonet” is not an exaggeration but a natural evolution: an internet that acts rather than informs.
And as a long‑time Transformers fan, I would be glad if we’d call these autonomous agents “Autobots.”
UX when the human is no longer the only user of the interface
As a UX designer, this raises a key question: who is the user in a future where as much traffic comes from autonomous agents as from humans?
Another colleague and Labs fellow, Jonas Hultenius, writes about this in his SogetiLabs blog post. That interfaces are designed for humans but already consumed by AI systems trying to interpret semantics, architecture, and signals never intended for them.
The autonet will require UX for AI; an AIX, or “AI‑as‑User” approach, where design no longer centers solely on human workflows. On the autonet, the agent’s experience must be as intentional as the human’s.
Delegation by default: when work continues without you
Fridlycke highlights a shift from “prompt → response → done” toward “delegate → supervise → review.” This mental model demands as much organizational and mindset transformation as technical change.
The user is no longer the primary executor. Systems carry the work forward over time. Our role shifts toward setting goals, boundaries, and quality standards, where the “workday” is no longer limited by human cognitive bandwidth. In the long run, I do see a scenario where humans may risk removing themselves from the loop entirely.
What do we want to achieve, and how?
The technology appears within reach. The real question is not if we can build an autonet, but how and why.
As UX designers, we must ask:
– Who needs this?
– What objectives does the solution serve?
– What goals should the agent pursue when the user isn’t watching?
– How do we express intention safely, clearly, and controllably?
– What happens to responsibility, trust, and transparency in a world where “the system did it for me”?
Sadly I don’t yet have the answers, only more questions. How do you wish for your train ticket be purchased? How do you like to discover inspiration for your new kitchen, and how do you feel that you retain control of that huge purchase?
So, what argues against the autonet?
We may be facing the most fascinating evolution of the internet yet: from a global information network to an autonomous operating layer for everyday life. From surfing to delegation. From interaction to intention.
Just as we optimized the web for algorithms through SEO, we continue to build digital worlds for both humans and the agents representing them.
The only counter‑argument I see is user demand and expectation. Some users clearly want an autonet, even investing money and their own e‑security into beta versions. But does the majority agree? I was part of the metaverse hype and saw the obvious potential in moving from flat 2D screens into an immersive VR experience. The masses however, did not.
What future awaits an autonomous internet remains to be seen.