This ER is suffering from the worst bout of lockdown since the COVID-19 pandemic. To prevent a ransomware attack, all online comms have been shut down in The Pitt season 2, meaning staff don’t have access to any resources or patient information.

While Dr. Baran Al-Hashimi (Sepideh Moafi) has been keen to up the AI ante since episode 2, efforts have spectacularly blown up in her face. The team are back to using whiteboards, pen and paper… but really, it’s only the tip of a chaotic iceberg.

The AI coding assistant Cursor has surpassed $2 billion in annualized revenue, according to a Bloomberg source. This individual says the four-year-old startup saw its revenue run rate double over the past three months.

The disclosure appears timed to counter a recent wave of skepticism. Last week, tweets went viral questioning whether Cursor’s momentum was stalling, citing high-profile defections by individual developers to competing tools — particularly Anthropic’s Claude Code.

Founded in 2022, Cursor initially sold its product primarily to individual developers. Over the last year, however, it has focused more on landing large corporate buyers, which now account for approximately 60% of revenue, according to Bloomberg.

While some individual developers and smaller startups have switched from Cursor to Claude Code, which is seen as more competitively priced, that attrition seems to higher-spending corporate customers who tend to stick around longer.

Beyond Claude Code, OpenAI’s coding tool Codex is also competing for share in the rapidly growing market for AI-assisted software development. Other startups in the space include Replit, Cognition, and Lovable.

Cursor was last valued at $29.3 billion when it raised a $2.3 billion funding round co-led by Accel and Coatue in November.

Cursor did not immediately respond to our request for comment.

Experts from Accenture, BearingPoint and Workhuman discuss how AI and automation can positively impact working life.

For many professionals artificial intelligence (AI) and automation have the power to transform day-to-day work. David Burke, a senior director of global talent acquisition and the employer brand at Workhuman explained that this transformation is effective not because it is ‘futuristic’ but because it meets the needs of an evolving workforce. 

“It’s much more practical than that,” he told SiliconRepublic.com. “We’re using AI across our internal systems to reduce manual work, improve decision-making and help teams move faster. The goal isn’t to replace roles, it’s to remove friction.

“In areas like hiring, performance enablement and cross-functional collaboration, automation is taking care of the repeatable tasks and surfacing better data. That means our teams spend less time chasing information or managing processes and more time solving problems and focusing on work that actually moves the business forward.”

This is a viewpoint shared by Wendy Walsh, a talent and organisation lead at Accenture, who noted that AI and automation have not only altered the tools she uses in the workplace, but have actually changed how she personally “shows up at work”.  

She said: “On a very practical level, I use AI every single day to think better. I use it to explore ideas, challenge my own assumptions, shape early thinking and get to a stronger point of view before anything ever becomes a document. For me, it’s less about productivity shortcuts and much more about cognitive support.”

Walsh added: “It helps me move faster to insight and clarity, not simply faster to output. The biggest difference is that my time has shifted away from preparing information and towards interpreting it.”

For BearingPoint’s Barry Haycock, who is a senior manager of data analytics and AI, when it comes to the topic of AI and automation, one subject that has dominated the conversation is agentic AI. He explained, he has noticed in the last 12 months or so, more and more people are choosing to use AI as an augmentation tool as opposed to automation.

He said: “In my personal day to day, I use AI to draft code I plan to write, or as a sounding board to discuss and tease out ideas before I start developing a slide deck or a document. 

“In many areas, people can use AI to perform a detailed search, for example of in-house documents, or to summarise their upcoming week and help them plan their goals. I find it useful too for flagging upcoming deadlines and prioritising them every Monday.”

For Walsh, Burke and Haycock, amid the evolution brought about by the proliferation of advanced technologies and processes in the working ecosystem, comes the need for a modern upskilling strategy. 

New day, new challenges

For Walsh, soft skills have grown in importance, with AI acting as a core reason why. She said: “As AI becomes part of everyday work, the qualities that really differentiate people are human ones. Skills in AI and data are important and technological literacy will increasingly be expected of everyone. But they’re not enough on their own. 

“Looking ahead to 2030, many of the fastest-growing core skills are deeply human. AI can analyse, generate and optimise at incredible speed. But it can’t build trust. It can’t create belonging. 

“It can’t decide what matters most in a moment of uncertainty. Technology is a powerful enabler, but it still needs people to shape it, question it and use it with purpose. The organisations that thrive will be those that invest just as seriously in human capability as they do in AI.”

Haycock said that in software development, MLOps and AIOps roles, business analytics is becoming the most important skill. He explained that, while the latest frontier AI models are excellent at coding or creating a script that a developer might need, the developer really needs to explain what’s required clearly. 

He said: “This is traditionally considered a soft skill and in times gone by a developer might write the code to explain their thoughts. I’ve noticed that ‘explain-in-plain-language’ skills are developing across many technical roles lately.”

“Technical skills will always matter,” said Burke, “but they’re increasingly learnable at speed. AI can help people acquire knowledge and capabilities faster than ever. What’s harder to automate and therefore more valuable, are human skills.” 

Skills such as judgement, communication, the ability to trust, context-setting, ethical decision-making and leading through ambiguity are among those that should be prioritised, especially as professionals are further expected to adopt and understand tech advancements. 

“As technology accelerates, the differentiator won’t be who knows the most,” said Burke. “It will be who can interpret, connect and lead. The irony is that the more advanced AI becomes, the more deeply human capability becomes alongside it. That’s what ultimately drives sustained performance.”

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

• The Xiaomi Watch 5 is now available for sale outside China
• The smartwatch originally launched in China in late 2025
• It’s switched operating systems from HyperOS to Google’s Wear OS 6

Back in December, Xiaomi launched its Watch 5 wearable with Wear OS 6 compatibility. The bad news is that it was only released in China, but the company has now made the device available worldwide, bringing its new features to a much wider audience.

The Watch 5 comes with a 1.54-inch AMOLED display featuring sapphire glass on both the front and back and reaching up to 1,500 nits of brightness. The body is made from a stainless steel frame and is available in Juniper Green or Black, while the 3D-printed strap weighs 43g, which Xiaomi says is about 50% lighter than a stainless steel equivalent.

Unlike on Earth there aren’t dozens of satellites whizzing around Mars to provide satellite navigation functionality. Recently NASA’s JPL engineers tried something with the Perseverance Mars rover that can give such Marsbound vehicles the equivalent of launching GPS satellites into Mars orbit, by introducing Mars Global Localization.

Although its remote operators back on Earth have the means to tell the rover where it is, it’d be incredibly helpful if it could determine this autonomously so that the rover doesn’t have to constantly stop and ask its human operators for directions. To this end the processor which was originally used to communicate with its Ingenuity helicopter companion was repurposed, reprogrammed to run an algorithm that compares panoramic images from the rover’s navigation cameras with its onboard orbital terrain maps.

Much like terrain-based navigation as used in cruise missiles back on Earth, this can provide excellent results depending on how accurate your terrain maps are. This terrain mapping process used to be done back on Earth, but for the past years engineers have worked to give the rover its own means to perform this task.

Because the off-the-shelf processor in the rover’s Helicopter Base Station (HBS) is much faster than the custom, radiation-hardened processors that control the rover, the decision was made to try the algorithm on the HBS, especially since Ingenuity was left behind after it fatally damaged its propeller during a rough landing. This left the HBS unused and free to be repurposed.

Repurposing such OTS hardware also provided a good way to check for radiation damage to such standard hardware that was never certified for high radiation environments. To validate reliability the algorithm was run multiple times on the HBS, with the results compared by the main computer. This found some discrepancies, attributed to damage to about 25 bits out of 1 GB of RAM.

By isolating these damaged bits, the algorithm could run reliably, while giving another nod to the genius of the Ingenuity program that enabled such new features with what was at the time an unproven and relatively low-budget side-project tacked onto the Perseverance rover.

Thanks to [Nevyn] for the tip.