Can AI Lead, Design, Govern—and Still Get the Details Right?

Linda Yaccarino is out as CEO of X, capping a turbulent two-year run after Elon Musk’s takeover of Twitter.

Her departure comes as the platform grapples with AI controversy, falling ad revenue, and rivals like Threads gaining ground.

She was the business brain to Musk’s product vision, but the Everything App dream just lost its lead navigator.

The UK just handed Google a golden opportunity: access to public sector infrastructure in exchange for free tech and AI training.

But critics say what looks like a smart deal now could trap the government in a future of Big Tech dependency.

With no tender and rising concerns about data sovereignty, privacy, and political influence, some see this not as progress, but a power handoff.

Blok, a new startup backed by $7.5M in funding, helps app developers simulate real-world user behavior using AI personas before launching features.

Founded by Tom Charman and Olivia Higgs, Blok allows teams to predict how different users might interact with a design based on event data and design files, offering early insights without releasing live betas.

With early traction in finance and healthcare, the company hopes to replace gut-driven A/B testing with predictive simulations.

Perplexity has launched Comet, an AI-powered web browser that integrates its search engine and a virtual assistant to automate tasks like summarizing emails and navigating webpages.

Initially available to Max subscribers, Comet represents Perplexity’s boldest move yet to rival Google Search and Chrome.

While the assistant impresses with simple tasks, it falters on complex ones and requires extensive access to user data. Still, Comet marks a notable step in the evolving AI-browser landscape.

In a power move, Replit is teaming up with Microsoft to land in Azure’s marketplace and integrate with core cloud services.

That’s a win for both — and a quiet blow to Google Cloud, where Replit apps are usually hosted. While the deal isn’t exclusive, it signals a broader shift in the cloud wars.

Replit’s rise is fast, its toolkit is flexible, and now, it’s playing in two big sandboxes.

Tencent has launched ArtifactsBench, a new benchmark designed to evaluate creative AI models based not just on code functionality but on user experience and design quality.

It uses an automated pipeline — from sandbox testing to a multimodal LLM judge — to score results on aesthetics, interactivity, and performance. The benchmark showed 94.4% alignment with human evaluations, significantly outperforming older methods.

Surprisingly, generalist AI models outperformed code-specific ones, pointing to the growing strength of all-purpose AI in creative tasks.