The No-Code Revolution Gets Real Desktop Apps

The No-Code Revolution Gets Real Desktop Apps

Raycast just launched Glaze, and it's the kind of tool that makes you rethink what "building software" means in 2026 [4]. Chat with AI, get a real desktop app in minutes—not a web wrapper, but actual native applications that live in your dock, work offline, and integrate with your system [4][5]. Powered by Claude Code and Codex, it's turning app creation into conversation.

The developer community is buzzing because this isn't another no-code toy—it's rapid prototyping that produces real software [6]. When you can go from idea to working desktop app faster than most people write documentation, the economics of custom software development just shifted dramatically.

This is what the post-code era looks like: AI handling implementation while humans focus on what the software should actually do. The question isn't whether this will disrupt traditional development—it's how fast.

OpenAI Brings Codex to Windows with Secure Agent Sandbox

OpenAI's Codex app is now fully available on Windows, complete with a native agent sandbox that uses PowerShell and OS-level security controls [7][8]. The sandbox employs restricted tokens, filesystem ACLs, and network restrictions to let AI agents run safely in developer environments, with WSL support for Linux workflows [7].

This matters because secure agent execution has been the missing piece for AI-assisted development at scale. When developers can trust AI agents to modify code, manage files, and execute commands without risking their systems, the productivity gains compound quickly.

The Windows release signals OpenAI's push beyond the browser into native development environments where serious work happens. As one developer noted, this expansion to Windows power users could accelerate enterprise adoption significantly [8].

24-Year-Old Ex-OpenAI Researcher Turns AI Thesis Into $5.5B Fund

Leopold Aschenbrenner—the researcher OpenAI fired for his "Situational Awareness" essay on AI safety—just proved his point with his wallet. His hedge fund grew from $225M to $5.5B in under 12 months, delivering 47% net returns by betting on AI infrastructure bottlenecks [9][10]. That's 8x the S&P 500's performance.

His strategy reads like a playbook for the AI boom's second-order effects: $885M in Bloom Energy fuel cells, positions in CoreWeave, bitcoin miners pivoting to AI compute, and shorts on traditional IT services like Infosys [9][11]. While everyone was buying AI software companies, Aschenbrenner was buying the power and compute that make them possible.

The irony is perfect: fired for warning about AI's rapid advancement, he made billions betting on exactly that acceleration. His success validates the thesis that AI's biggest constraint isn't algorithms—it's electricity and silicon.

Finland Leads EU in AI Act Enforcement

Finland became the first EU member state to fully operationalize the AI Act on January 1st, with Traficom serving as the single enforcement point [12][13]. The new laws grant authorities power to inspect systems, demand documentation, halt non-compliant AI deployments, and levy fines [12].

This positions Finland as the EU's AI regulation laboratory while other member states are still figuring out their enforcement frameworks [14]. For companies building AI products in Europe, Finland just became the canary in the regulatory coal mine—what works there will likely spread across the bloc.