In a new InfoQ podcast, UX veteran Scott Jenson argues that desktop operating systems have stagnated for decades, cloud-first software has traded convenience for fragility, and the next wave of AI will be more useful when it is embedded in workflows rather than wrapped in a universal chatbot. Speaking with InfoQ’s Olimpiu Pop, the designer behind work on Macintosh System 7, Google Maps, Chrome and other products made the case for more local-first, context-aware computing in an era he believes is quickly moving toward “agentic” UX.
Scott Jenson’s case against the status quo
Jenson said the desktop has not fundamentally evolved in about 20 years. In his view, the main changes from major platform vendors have largely strengthened ecosystems rather than improved user productivity. He pointed to features such as Universal Clipboard, AirDrop and phone-to-desktop integration as examples of tighter platform control, not a rethinking of what desktop software can do.
His broader critique is that the industry has become too comfortable with existing interface patterns. Rather than trying new approaches, software makers often preserve familiar models and optimize around lock-in, he said.
Why mobile did not replace desktop
Pop raised the long-running debate over whether tablets and phones would eventually displace desktop systems. Jenson pushed back, saying mobile “won consumers” but not productivity. Tablets, he argued, are optimized for consumption, gaming and social apps, not the kind of creation-heavy work that still depends on a desktop.
He described a failed internal experiment at Google that tried to push tablet use at work around 2015. The result, he said, showed that tablets are useful for a narrow set of tasks, but not a replacement for desktop workflows.
- Mobile is strong for lightweight, focused tasks.
- Tablets fit content and entertainment use cases well.
- Desktops still matter for complex, creation-oriented work.
Local-First and the limits of cloud-centric software
The conversation then turned to Local-First architecture, which Jenson sees as a practical response to cloud dependency. He said the cloud raises two major concerns: privacy, because user data often sits with a single provider, and resilience, because work can be disrupted by connectivity problems or provider outages.
He framed Local-First as a way to keep data under the user’s control while still taking advantage of modern services. The goal is not to reject cloud software outright, but to reduce fragility and let people continue working even when network access fails.
Pop brought up examples such as large phone libraries and low-quality previews that still force users back to the cloud. Jenson’s response was that the current model often makes data harder to work with, not easier, especially on mobile systems that hide files inside individual apps.
Desktop OS as working memory
One of Jenson’s central themes was that operating systems should act more like working memory. He contrasted this with the direct manipulation model used by most modern interfaces, which he described as stateless: users cut, paste, open, close and move on, but the system does not remember much context.
In his view, future desktops should preserve more of what users were doing so they can resume tasks more naturally after interruptions. That might mean keeping project context visible, capturing activity over time or exposing richer history than today’s clipboard and file browser patterns allow.
Three prototypes for a more contextual desktop
Jenson outlined three desktop prototypes he presented at a Local-First conference, each aimed at creating a kind of working memory for users.
- Spatial model: a wide, more organic window manager that spreads activity across a larger visual field instead of rigid virtual desktops.
- Associative clipboard/project drawer: a side panel that holds clipboard history, files and related work items for a specific task or project.
- Episodic timeline: a ribbon-like history of what the user has done during the day, inspired by the 1990s Lifestreams research.
He said the spatial approach is meant to be less rigid than virtual desktops, which he thinks can work well for highly organized users but feel awkward for everyone else. The clipboard and timeline ideas, by contrast, are about helping people recover context and find information by association or by time.
Jenson stressed that these are prototypes, not finished answers. Still, he believes they show that desktop operating systems have room to become more helpful without being torn down and rebuilt from scratch.
Agentic UX: beyond the chatbot phase
The discussion also covered what Jenson called the future of LLM integration. He argued that chatbots are an easy starting point, but not the final shape of AI in software. Instead of forcing users to talk to a general-purpose assistant for everything, he expects intelligence to move into the application itself and become part of the workflow.
He described a recent example in which a model explained what a pivot table was and then, in a more advanced interaction, built one inside Google Sheets. That distinction matters to him: chat can teach, but the next step is doing the work inside the app.
Jenson also cautioned against assuming that adding more skills to a chatbot is the best route. He pointed to products like Napkin AI as an example of systems that use conversation to produce intermediate artifacts and expose options through conventional UX elements rather than relying solely on a chat interface.
Designers, coding agents and faster iteration
For Jenson, one of the most promising uses of agentic tools is speeding up design iteration. He said UX designers can use coding agents to prototype ideas faster, test variations more quickly and reach higher-quality results through experimentation.
He emphasized that this does not replace programmers. Instead, it lets designers work at a much higher pace and explore interface concepts that would otherwise take too long to build manually. He is also turning some of his prototypes into actual code, including a timeline project he has implemented as both a Chrome extension and a Swift extension.
Standards, regulation and the role of government
Jenson and Pop also discussed standards and regulation, including the European Union’s influence on Apple’s move from Lightning to USB-C. Jenson said that sort of pressure can be useful when platform companies do not have users’ long-term interests in mind.
He framed the issue less as a standards debate and more as compliance and public-interest regulation. In his view, governments have a legitimate role in pushing companies toward longer-term thinking, especially when business incentives favor short-term control and ecosystem lock-in.
He also tied the conversation back to climate and resource use, saying short-term decision-making has broader social costs that companies are unlikely to account for on their own.
A practical, not destructive, vision for the future
Despite the sharp criticism, Jenson repeatedly said he is not trying to blow up existing systems. His goal is to make the problems more visible and encourage experimentation. He thinks both desktop and mobile have been frozen in time, and that the industry needs to explore alternatives rather than assume today’s patterns are permanent.
He closed by saying the shift toward local models, agentic tools and more contextual interfaces may eventually move power away from the largest platform companies. But for that to happen, software has to move beyond the “status quo” and become more supportive of how people actually think and work.
Source: Original report
Was this helpful?
Explore more: AI Automation Services More AI & Automation Tech News
Last Modified: September 2, 2026 at 1:53 am
3 views
